The goal of Scottish-cycle-counts is to read-in a process data on cycling volumes in Scotland.
library(tidyverse)First of all, a copy of the original files can be obtained with the code below. This code is to be run just once.
dir.create("data_raw")
system("gh release download 1 --dir data_raw")The input dataset is a single .zip file:
zipped_data = list.files(path = "data_raw",pattern = "\\.zip$",full.names = T)
zipped_dataWe can unzip it as follows:
unzip(zipped_data, exdir = "data_raw")files_csv = list.files("data_raw", pattern = "\\.csv$", full.names = TRUE)
files_csv
#> [1] "data_raw/Aberdeen City Council.csv"
#> [2] "data_raw/Aberdeenshire Council.csv"
#> [3] "data_raw/City of Edinburgh Council.csv"
#> [4] "data_raw/Comhairle nan Eilean Siar (Western Isles Council).csv"
#> [5] "data_raw/East Ayrshire Council.csv"
#> [6] "data_raw/East Dunbartonshire Council.csv"
#> [7] "data_raw/Glasgow City Council.csv"
#> [8] "data_raw/John Muir Way.csv"
#> [9] "data_raw/National Cycle Nework (Scotland) - Sustrans.csv"
#> [10] "data_raw/National Monitoring Framework - Cycling Scotland.csv"
#> [11] "data_raw/North Ayrshire Council.csv"
#> [12] "data_raw/Perth and Kinross Council.csv"
#> [13] "data_raw/Scotland North East Trunk Roads.csv"
#> [14] "data_raw/Scotland North West Trunk Roads.csv"
#> [15] "data_raw/Scotland South East Trunk Roads.csv"
#> [16] "data_raw/Scotland South West Trunk Roads.csv"
#> [17] "data_raw/South Ayrshire Council.csv"
#> [18] "data_raw/South Lanarkshire Council.csv"
#> [19] "data_raw/Stirling Council.csv"
#> [20] "data_raw/The Highland Council.csv"We can read this file in R as follows:
# library(data.table)
# counts = data.frame(data.table::rbindlist(lapply(files_csv,data.table::fread))) #DT's quick way to read the files
counts = map_dfr(files_csv, read_csv,show_col_types = FALSE)
dim(counts)
#> [1] 230307 10
counts
#> # A tibble: 230,307 × 10
#> area count endTime latitude location longitude provider siteID
#> <chr> <dbl> <dttm> <dbl> <chr> <dbl> <chr> <chr>
#> 1 Aberde… 61 2023-06-14 00:00:00 57.1 A956 We… -2.09 Aberdee… ABE65…
#> 2 Aberde… 26 2023-06-14 00:00:00 57.2 Dyce Dr… -2.20 Aberdee… ABE918
#> 3 Aberde… 18 2023-06-14 00:00:00 57.2 Dyce Dr… -2.22 Aberdee… ABE397
#> 4 Aberde… 158 2023-06-14 00:00:00 57.2 Ellon R… -2.09 Aberdee… ABE920
#> 5 Aberde… 157 2023-06-14 00:00:00 57.1 Shell C… -2.10 Aberdee… ABE229
#> 6 Aberde… 132 2023-06-14 00:00:00 57.2 Tillydr… -2.11 Aberdee… ABE899
#> 7 Aberde… 395 2023-06-14 00:00:00 57.1 Deeside… -2.10 Aberdee… ABE235
#> 8 Aberde… 20 2023-06-14 00:00:00 57.2 Farburn… -2.17 Aberdee… ABE745
#> 9 Aberde… 91 2023-06-14 00:00:00 57.2 Parkway… -2.10 Aberdee… ABE894
#> 10 Aberde… 39 2023-06-14 00:00:00 57.2 Site B … -2.16 Aberdee… ABE288
#> # ℹ 230,297 more rows
#> # ℹ 2 more variables: startTime <dttm>, usmart_id <chr>counts_monthly = counts |>
mutate(
year = year(endTime),
month = month(endTime, label = TRUE),
# date rounded to nearest month in 2020-01-01 format:
date = lubridate::floor_date(endTime, unit = "month")
) |>
group_by(date, area) |>
summarise(
count = sum(count)
)
# Add column with names for most common areas:
# Most common areas are:
area_counts = counts_monthly |>
group_by(area) |>
summarise(
count = sum(count)
)
top_5_areas = area_counts |>
slice_max(count, n = 5)
# Add column that is area name if in top 5, else "Other":
counts_monthly_top = counts_monthly |>
mutate(
Area = ifelse(
area %in% top_5_areas$area,
area,
"Other"
)
) |>
group_by(date, Area) |>
summarise(
count = sum(count)
)counts_monthly_top |>
ggplot(aes(x = date, y = count, colour = Area)) +
geom_line() +
# Add log y-axis:
scale_y_log10()
A quick look at the data to check if there are sites with missing data or duplicated records:
counts |>
group_by(siteID,location) |>
unique() |>
count() |>
ggplot(aes(n))+geom_histogram(bins = 30)
range(counts$startTime)
#> [1] "2022-06-01 UTC" "2023-06-13 UTC"
difftime(range(counts$startTime)[2],range(counts$startTime)[1],units = "days")
#> Time difference of 377 daysEach site should have a maximum of 378 days in the dataset. The following code detects the sites with some type of duplication and the ones with fewer records.
repeated_sites = counts |>
group_by(siteID) |>
filter(n()>378) |>
select(siteID) |>
unique()
repeated_sites$siteID |> length()
#> [1] 22
fewer_sites = counts |>
group_by(siteID) |>
filter(n()<300) |>
select(siteID) |>
unique()
fewer_sites$siteID |> length()
#> [1] 133A subset of the clean sites is produced, so we can do some AADT analysis. Records after 2023-06-01 are filtered out to have only one year of data for each site
clean_counts = counts |>
filter(startTime < as.Date("2023-06-01")) |>
anti_join(repeated_sites,by =join_by(siteID)) |>
anti_join(fewer_sites,by =join_by(siteID)) |>
filter(n()==365,.by = siteID)
clean_counts |>
group_by(siteID,location) |>
unique() |>
count() |> summary()
#> siteID location n
#> Length:383 Length:383 Min. :365
#> Class :character Class :character 1st Qu.:365
#> Mode :character Mode :character Median :365
#> Mean :365
#> 3rd Qu.:365
#> Max. :365Here we calculate some statistics for the whole year including mean (Average Annual Daily Flow), median daily flow, minimum and maximum daily flows,
AADF_sites = clean_counts |>
summarise(across(count,list(mean = mean,
median = median,
min = min,
max = max)),.by = siteID)
AADF_sites
#> # A tibble: 383 × 5
#> siteID count_mean count_median count_min count_max
#> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 ABE6535 19.9 18 1 50
#> 2 ABE896 44.1 43 0 195
#> 3 ABE6545 34.3 32 2 84
#> 4 ABE895 86.0 74 0 288
#> 5 ABE918 15.6 15 0 39
#> 6 ABE122 16.6 16 0 57
#> 7 ABE123 13.8 12 0 49
#> 8 ABE400 43.5 33 2 150
#> 9 ABE288 12.5 9 0 81
#> 10 ABE893 43.3 43 0 93
#> # ℹ 373 more rowscounts_per_area = counts |> select(siteID,area) |> unique()
AADF_sites |>
left_join(counts_per_area,by = join_by(siteID)) |>
mutate(
area = ifelse(
area %in% top_5_areas$area,
area,
"Other"
)) |>
ggplot(aes(x=fct_reorder(area,count_mean,.desc = T),
y=count_mean))+
geom_boxplot(outlier.shape = NA)+
geom_jitter(aes(col = area),alpha = 0.2,shape = 19,show.legend = F)+
coord_cartesian(ylim = c(0,1000))+
theme_minimal()+
labs(x = "Area",
y = "AADF")
We create a vector to store the bank holidays in Scotland extracted from the mygov.scot web
scot_bank_holidays = as.Date(c("2022/06/03",
"2022/06/02",
"2022/08/1",
"2022/11/30",
"2022/12/25",
"2022/12/26",
"2022/12/27",
"2023/01/02",
"2023/01/03",
"2023/04/07",
"2023/05/01",
"2023/05/08",
"2023/05/29"))We can calculate the same summary statistics by type of day: bank holidays, weekends and weekdays (AAWDF).
ADF_dtype = clean_counts |>
mutate(d.type = case_when(as.Date(startTime) %in% scot_bank_holidays~"Bank Holiday",
wday(startTime,week_start = 1)<6~"Weekday",
TRUE~"Weekend")) |>
summarise(across(count,list(mean = mean,
median = median,
min = min,
max = max)),
.by = c(siteID,d.type))
ADF_dtype
#> # A tibble: 1,149 × 6
#> siteID d.type count_mean count_median count_min count_max
#> <chr> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 ABE6535 Weekday 23.6 22 6 50
#> 2 ABE896 Weekday 46.4 46 2 101
#> 3 ABE6545 Weekday 40.3 40 2 84
#> 4 ABE895 Weekday 78.1 65 1 247
#> 5 ABE918 Weekday 19.5 20 0 39
#> 6 ABE122 Weekday 17 16 0 50
#> 7 ABE123 Weekday 12.7 11 0 45
#> 8 ABE400 Weekday 40.1 31 2 110
#> 9 ABE288 Weekday 11.8 8 0 62
#> 10 ABE893 Weekday 52.6 52 6 93
#> # ℹ 1,139 more rowscounts_per_area = counts |> select(siteID,area) |> unique()
ADF_dtype |>
left_join(counts_per_area,by = join_by(siteID)) |>
mutate(
area = ifelse(
area %in% top_5_areas$area,
area,
"Other"
)) |>
ggplot(aes(x=fct_reorder(area,count_mean,.desc = T),
y=count_mean))+
geom_boxplot(outlier.shape = NA)+
geom_jitter(aes(col = area),alpha = 0.2,shape = 19,show.legend = F)+
coord_cartesian(ylim = c(0,1000))+
facet_wrap(d.type~.,ncol = 2)+
theme_minimal()+
labs(x = "Area",
y = "ADF")
library(sf)
library(tmap)The following code reads the network with the estimated commute trips
from the npt repository. Each edge/link of the network has four
attributes. We will focus on bicycle which is the estimated number of
daily commute trips in both directions.
rnet_commute = read_rds("../npt/outputs/rnet_commute.Rds")
rnet_commute
#> Simple feature collection with 1709 features and 4 fields
#> Geometry type: LINESTRING
#> Dimension: XY
#> Bounding box: xmin: -3.33653 ymin: 55.89548 xmax: -3.12415 ymax: 55.98413
#> Geodetic CRS: WGS 84
#> First 10 features:
#> bicycle bicycle_go_dutch Gradient Quietness geometry
#> 1 0 1 0 40 LINESTRING (-3.1547 55.9749...
#> 2 0 1 0 60 LINESTRING (-3.27797 55.911...
#> 3 0 1 2 70 LINESTRING (-3.27672 55.910...
#> 4 0 1 4 70 LINESTRING (-3.14531 55.969...
#> 5 0 1 0 80 LINESTRING (-3.27741 55.911...
#> 6 0 1 0 80 LINESTRING (-3.15282 55.976...
#> 7 0 1 1 80 LINESTRING (-3.17113 55.961...
#> 8 0 1 0 90 LINESTRING (-3.1547 55.9749...
#> 9 0 1 1 90 LINESTRING (-3.16781 55.961...
#> 10 0 1 1 90 LINESTRING (-3.16555 55.962...An sf object is created from the clean_counts data frame. AADF for
each counts are joined using the siteID
sf_counts = clean_counts |>
select(siteID,latitude,longitude,provider,location) |>
unique() |>
left_join(AADF_sites,by = "siteID") |>
filter(count_mean > 0) |>
st_as_sf(coords = c("longitude","latitude"),crs = 4326)
sf_counts
#> Simple feature collection with 338 features and 7 fields
#> Geometry type: POINT
#> Dimension: XY
#> Bounding box: xmin: -7.307251 ymin: 54.91291 xmax: -1.15021 ymax: 60.1511
#> Geodetic CRS: WGS 84
#> # A tibble: 338 × 8
#> siteID provider location count_mean count_median count_min count_max
#> * <chr> <chr> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 ABE6535 Aberdeen City C… A96 Auc… 19.9 18 1 50
#> 2 ABE896 Aberdeen City C… F&B Way… 44.1 43 0 195
#> 3 ABE6545 Aberdeen City C… A9119 Q… 34.3 32 2 84
#> 4 ABE895 Aberdeen City C… Beach E… 86.0 74 0 288
#> 5 ABE918 Aberdeen City C… Dyce Dr… 15.6 15 0 39
#> 6 ABE122 Aberdeen City C… Maidenc… 16.6 16 0 57
#> 7 ABE123 Aberdeen City C… Maidenc… 13.8 12 0 49
#> 8 ABE400 Aberdeen City C… Seaton … 43.5 33 2 150
#> 9 ABE288 Aberdeen City C… Site B … 12.5 9 0 81
#> 10 ABE893 Aberdeen City C… Welling… 43.3 43 0 93
#> # ℹ 328 more rows
#> # ℹ 1 more variable: geometry <POINT [°]>A subset of the counts are taken based on a buffer of the rnet_commute
object.
rnet_buffer20 = rnet_commute |> st_union() |> st_buffer(dist = 20)
sf_counts_selected = sf_counts[rnet_buffer20,]The nearest feature is joined to each point location
val_app1 = st_join(sf_counts_selected,rnet_commute,join = st_nearest_feature)
val_app1
#> Simple feature collection with 21 features and 11 fields
#> Geometry type: POINT
#> Dimension: XY
#> Bounding box: xmin: -3.24015 ymin: 55.92337 xmax: -3.12779 ymax: 55.9566
#> Geodetic CRS: WGS 84
#> # A tibble: 21 × 12
#> siteID provider location count_mean count_median count_min count_max
#> * <chr> <chr> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 EDH0030 City of Edinbur… Dalry R… 1009. 331 40 27670
#> 2 EDH0037 City of Edinbur… Whiteho… 437. 457 54 753
#> 3 EDH0045 City of Edinbur… Melvill… 128. 128 13 266
#> 4 EDH0022 City of Edinbur… North M… 885. 888 99 1839
#> 5 EDH0035 City of Edinbur… A90 Dea… 415. 361 36 857
#> 6 EDH0039 City of Edinbur… Bruntsf… 218. 236 0 481
#> 7 EDH0043 City of Edinbur… Melvill… 1099. 1109 242 1813
#> 8 EDH0041 City of Edinbur… Mayfiel… 152. 0 0 683
#> 9 EDH0029 City of Edinbur… A8 Cors… 100. 0 0 829
#> 10 EDH0044 City of Edinbur… Melvill… 251. 257 32 446
#> # ℹ 11 more rows
#> # ℹ 5 more variables: geometry <POINT [°]>, bicycle <dbl>,
#> # bicycle_go_dutch <dbl>, Gradient <dbl>, Quietness <dbl>tmap_mode("view")
tm_shape(rnet_commute)+
tm_lines(col = "bicycle",lwd = "bicycle",lwd.legend = tm_legend_combine("col"))+
tm_shape(val_app1)+
tm_dots(col = "count_mean")The following plot compares the observed counts and the paired estimate flows.
val_app1 |>
st_drop_geometry() |>
mutate(ratio = bicycle/count_mean) |>
ggplot(aes(ratio))+
geom_histogram()
val_app1 |>
st_drop_geometry() |>
ggplot(aes(x = count_mean,
y = bicycle)) +
geom_point() +
geom_smooth(method = "lm",
formula = 'y ~ x',
se = F) +
coord_fixed(xlim = c(0, max(
c(val_app1$count_mean, val_app1$bicycle)
)),
ylim = c(0, max(
c(val_app1$count_mean, val_app1$bicycle)
)))
A linear regression is used to evaluate the fit of the estimates, it is
assumed that the proportion of commute trips is constant across counts
(intercept = 0).
lm_app1 = lm(bicycle ~ count_mean+0,data = val_app1)
summary(lm_app1)
#>
#> Call:
#> lm(formula = bicycle ~ count_mean + 0, data = val_app1)
#>
#> Residuals:
#> Min 1Q Median 3Q Max
#> -4.8149 -0.5083 0.6479 4.4314 8.0276
#>
#> Coefficients:
#> Estimate Std. Error t value Pr(>|t|)
#> count_mean 0.006192 0.001606 3.855 0.000988 ***
#> ---
#> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#>
#> Residual standard error: 4.054 on 20 degrees of freedom
#> Multiple R-squared: 0.4263, Adjusted R-squared: 0.3976
#> F-statistic: 14.86 on 1 and 20 DF, p-value: 0.0009876The previous approach assigned a count site to each road link based. So far, it has not been addressed the fact that counts might be reporting uni-directional flows along specific links.
Using the sf_counts_selected object, we produce a buffer of 30 metres
from each count site.
sel_counts_buf30 = st_buffer(sf_counts_selected,dist = 30)All overlaps within the sel_counts_buf30 are identified using the
following code.
counts_overlap = st_intersects(sel_counts_buf30, sel_counts_buf30)Some overlaps might reveal count sites reporting flows of different directions on the same edge/link. For example, the sites EDH0040 and EDH0041 which are located on Mayfield Road.
tmap_mode("view")
tm_shape(sel_counts_buf30[sel_counts_buf30$siteID %in% c("EDH0040","EDH0041"),])+
tm_polygons(alpha = 0.5)+
tm_shape(sf_counts_selected)+
tm_dots()
A more complex instance is the overlap of sensors on Melville Dr, two of the sensors report flows on the main road and the other has data of the adjacend path.
tmap_mode("view")
tm_shape(sel_counts_buf30[sel_counts_buf30$siteID %in% c("EDH0042","EDH0043","EDH0044","EDH0045"),])+
tm_polygons(alpha=0.3)+
tm_shape(sf_counts_selected)+
tm_dots()The following code aggregates some of the overlapping counts using the
location attribute as aggregation criteria.
grouped_counts =
do.call(rbind,
lapply(unique(counts_overlap),
function(x) {
tmp_group = sf_counts_selected[x,]
# Count aggreagation
simp_data = tmp_group |>
# Removing the direction from the location string
mutate(location = str_remove(location,
"\\s\\w*bound")) |>
st_drop_geometry() |>
# Extracting the first value for the siteID,
# provider and adds up the counts for sites with
# the same 'location'
summarise(across(c("siteID", "provider"),
\(x) head(x, n = 1)),
across(starts_with("count_"), sum),
.by = "location")
simp_group = tmp_group |>
select(siteID) |>
filter(siteID %in% simp_data$siteID)
simp_counts = simp_group |>
left_join(simp_data, by = "siteID") |>
relocate(location, .after = provider) |>
relocate(geometry, .after = count_max)
return(simp_counts)
}))As in the previous approach, the count sites are joined to the nearest
feature in the rnet_commute network.
grouped_counts$nearest_edge = st_nearest_feature(grouped_counts,
rnet_commute,
check_crs = T)
val_app2 = cbind(grouped_counts,st_drop_geometry(rnet_commute)[grouped_counts$nearest_edge,])The following code shows the counts with the corresponding network edges/links.
tm_shape(grouped_counts)+
tm_dots()+
tm_shape(rnet_commute[grouped_counts$nearest_edge,])+
tm_lines()The figure below compares the counts and the estimated flows for the current approach
val_app2 |>
st_drop_geometry() |>
ggplot(aes(x = count_mean,
y = bicycle)) +
geom_point() +
geom_smooth(method = "lm",
formula = 'y ~ x',
se = F) +
coord_fixed(xlim = c(0, max(
c(val_app2$count_mean, val_app2$bicycle)
)),
ylim = c(0, max(
c(val_app2$count_mean, val_app2$bicycle)
)))
As in the previous approach, a linear regression is used to have a high-level assessment of the estimations. With this approach, although there is not a significant change in the estimate, there is a slight improvement in the R2.
lm_app2 = lm(bicycle ~ count_mean+0,data = val_app2)
summary(lm_app2)
#>
#> Call:
#> lm(formula = bicycle ~ count_mean + 0, data = val_app2)
#>
#> Residuals:
#> Min 1Q Median 3Q Max
#> -4.6021 -0.7120 0.2573 4.3372 8.0632
#>
#> Coefficients:
#> Estimate Std. Error t value Pr(>|t|)
#> count_mean 0.005965 0.001523 3.917 0.00111 **
#> ---
#> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#>
#> Residual standard error: 3.916 on 17 degrees of freedom
#> Multiple R-squared: 0.4744, Adjusted R-squared: 0.4435
#> F-statistic: 15.35 on 1 and 17 DF, p-value: 0.001109Downloading the data from latest builds. RDS objects with the list of networks have been selected for this analysis
oldwd <- setwd("../npt/outputdata")
system("gh release download v2023-08-18-10-42-44_commit_cbb84b024550d638dbca066c5850d1b03d55fc66 --clobber")
system("gh release download v2023-09-10-17-43-21.109279_commit_86ae338b12f523c27fcc290f48105f2e5dbdcab7 --clobber")
setwd(oldwd)We select the ones that contain the naming pattern with a regular expression.
networks_files = list.files(path = "../npt/outputdata/",pattern = "^rnet.*_list",full.names = TRUE)
networks_files
#> [1] "../npt/outputdata/rnet_commute_list.Rds"
#> [2] "../npt/outputdata/rnet_school_list.Rds"The following function is prepared to run the analysis as per the second approach. Note that the linear model used for the evaluation does not include an intercept.
library(purrr)
rnet_val = function(rnet_path,counts){
# Loading the network RDS file
rnet_nested_list = read_rds(rnet_path)
# Detecting the name of the main purpose
main_rnet_name = str_extract(rnet_path,"(commute|school)")
# Flattening the list
rnet_flat_list = rnet_nested_list |> list_flatten()
lst_names = paste(main_rnet_name,names(rnet_flat_list),sep = ".")
# Assigning names to the lists
names(rnet_flat_list) = lst_names
lm_rnet =
lapply(lst_names,function(rnet_name){
rnet = rnet_flat_list[[rnet_name]]
# Creating buffer
rnet_buffer20 = rnet |>
st_union() |>
st_buffer(dist = 20)
# Subsetting counts based on buffer
counts_selected = counts[rnet_buffer20,]
# Creating buffer around counts 30 m
counts_buf30 = st_buffer(counts_selected,dist = 30)
# Finding overlapping counts
counts_overlap_30 = st_intersects(counts_buf30, counts_buf30)
# Processing overlaps and aggregating if possible
aggregated_counts =
do.call(rbind,
lapply(unique(counts_overlap_30),
function(x) {
tmp_group = counts_selected[x, ]
# Count aggreagation
simp_data = tmp_group |>
# Removing the direction from the location string
mutate(location = str_remove(location,
"\\s\\w*bound")) |>
st_drop_geometry() |>
# Extracting the first value for the siteID,
# provider and adds up the counts for sites with
# the same 'location'
summarise(across(c("siteID", "provider"),
\(x) head(x, n = 1)),
across(starts_with("count_"), sum),
.by = "location")
simp_group = tmp_group |>
select(siteID) |>
filter(siteID %in% simp_data$siteID)
simp_counts = simp_group |>
left_join(simp_data, by = "siteID") |>
relocate(location, .after = provider) |>
relocate(geometry, .after = count_max)
return(simp_counts)
}))
aggregated_counts$nearest_edge = st_nearest_feature(aggregated_counts,
rnet,
check_crs = T)
val_counts = cbind(aggregated_counts,
st_drop_geometry(rnet)[aggregated_counts$nearest_edge,])
lm_counts = lm(bicycle ~ count_mean + 0,data = val_counts)
return(lm_counts)
})
names(lm_rnet) = lst_names
return(lm_rnet)
}The following code runs the validation for each network
val_results = lapply(networks_files,
rnet_val,
counts = sf_counts
)Summarising the high-level results for all networks:
library(broom)
val_results_flat = val_results |> list_flatten(name_spec = "{inner}")
tbl_val_results = tibble(network = names(val_results_flat),
intercept = 0,
coef = vapply(val_results_flat,coef,FUN.VALUE = 0),
R2 = vapply(val_results_flat,\(x) summary(x)$adj.r.squared,FUN.VALUE = 0)
)
tbl_val_results
#> # A tibble: 12 × 4
#> network intercept coef R2
#> <chr> <dbl> <dbl> <dbl>
#> 1 commute.fastest 0 0.0626 0.0912
#> 2 commute.balanced 0 0.0895 0.0954
#> 3 commute.quietest 0 0.0953 0.0751
#> 4 commute.ebike 0 0.0553 0.0814
#> 5 school.Primary_fastest 0 0.00229 0.0643
#> 6 school.Primary_balanced 0 0.00217 0.0410
#> 7 school.Primary_quietest 0 0.00270 0.0459
#> 8 school.Primary_ebike 0 0.00185 0.0468
#> 9 school.Secondary_fastest 0 0.00220 0.0274
#> 10 school.Secondary_balanced 0 0.00263 0.0321
#> 11 school.Secondary_quietest 0 0.00250 0.0227
#> 12 school.Secondary_ebike 0 0.00202 0.0227rm(val_results,val_results_flat)Given that the proportion of trips by purpose might vary spatially, e.g. urban areas might have a higher proportion of commuting trips. A weighted geographical model is implemented. The default bi-square function is used to determine the kernel bandwidth.
library(GWmodel)
rnet_val_gwr = function(rnet_path, counts) {
# Loading the network RDS file
rnet_nested_list = read_rds(rnet_path)
# Detecting the name of the main purpose
main_rnet_name = str_extract(rnet_path, "(commute|school)")
# Flattening the list
rnet_flat_list = rnet_nested_list |> list_flatten()
lst_names = paste(main_rnet_name, names(rnet_flat_list), sep = ".")
# Assigning names to the lists
names(rnet_flat_list) = lst_names
reg_rnet =
lapply(lst_names, function(rnet_name) {
rnet = rnet_flat_list[[rnet_name]]
# Creating buffer
rnet_buffer20 = rnet |>
st_union() |>
st_buffer(dist = 20)
# Subsetting counts based on buffer
counts_selected = counts[rnet_buffer20, ]
# Creating buffer around counts 30 m
counts_buf30 = st_buffer(counts_selected, dist = 30)
# Finding overlapping counts
counts_overlap_30 = st_intersects(counts_buf30, counts_buf30)
# Processing overlaps and aggregating if possible
aggregated_counts =
do.call(rbind,
lapply(unique(counts_overlap_30),
function(x) {
tmp_group = counts_selected[x,]
# Count aggreagation
simp_data = tmp_group |>
# Removing the direction from the location string
mutate(location = str_remove(location,
"\\s\\w*bound")) |>
st_drop_geometry() |>
# Extracting the first value for the siteID,
# provider and adds up the counts for sites with
# the same 'location'
summarise(across(c("siteID", "provider"),
\(x) head(x, n = 1)),
across(starts_with("count_"), sum),
.by = "location")
simp_group = tmp_group |>
select(siteID) |>
filter(siteID %in% simp_data$siteID)
simp_counts = simp_group |>
left_join(simp_data, by = "siteID") |>
relocate(location, .after = provider) |>
relocate(geometry, .after = count_max)
return(simp_counts)
}))
aggregated_counts$nearest_edge = st_nearest_feature(aggregated_counts,
rnet,
check_crs = T)
val_counts = cbind(aggregated_counts,
st_drop_geometry(rnet)[aggregated_counts$nearest_edge, ])
bw <- bw.gwr(
formula = bicycle ~ count_mean,
approach = "AIC",
adaptive = T,
data = as_Spatial(val_counts)
)
gwr.mod <- gwr.basic(
formula = bicycle ~ count_mean,
adaptive = T,
data = as_Spatial(val_counts),
bw = bw
)
return(gwr.mod)
})
names(reg_rnet) = lst_names
return(reg_rnet)
}As in the previous attempt, the following code runs the validation using a GWR model.
Summarising the high-level results for all networks for the GWR model:
library(broom)
val_results_gwr_flat = val_results_gwr |>
list_flatten(name_spec = "{inner}")
tbl_val_gwr_results = tibble(network = names(val_results_gwr_flat),
mean_intercept = vapply(val_results_gwr_flat,\(x) mean(x$SDF@data[, 1]),FUN.VALUE = 0),
median_intercept = vapply(val_results_gwr_flat,\(x) median(x$SDF@data[, 1]),FUN.VALUE = 0),
mean_coef = vapply(val_results_gwr_flat,\(x) mean(x$SDF@data[, 2]),FUN.VALUE = 0),
median_coef = vapply(val_results_gwr_flat,\(x) median(x$SDF@data[, 2]),FUN.VALUE = 0),
bw = vapply(val_results_gwr_flat,\(x) x$GW.arguments$bw,FUN.VALUE = 0),
R2 = vapply(val_results_gwr_flat,\(x) x$GW.diagnostic$gw.R2,FUN.VALUE = 0)
)
tbl_val_gwr_results
#> # A tibble: 12 × 7
#> network mean_intercept median_intercept mean_coef median_coef bw R2
#> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 commute.f… 14.3 8.54 5.60e-2 0.0122 140 0.223
#> 2 commute.b… 21.8 17.6 7.87e-2 0.0259 150 0.204
#> 3 commute.q… 26.4 20.6 8.44e-2 0.0386 150 0.166
#> 4 commute.e… 15.2 8.70 4.49e-2 0.00896 140 0.209
#> 5 school.Pr… 1.44 1.47 3.78e-3 0.00148 40 0.258
#> 6 school.Pr… 1.98 1.89 -1.41e-4 -0.000296 136 0.0590
#> 7 school.Pr… 2.30 1.94 -9.89e-5 0.000105 135 0.0851
#> 8 school.Pr… 1.44 1.43 2.05e-4 0.00000476 163 0.0677
#> 9 school.Se… 1.44 0.772 -4.50e-3 0.000617 31 0.548
#> 10 school.Se… 1.59 0.966 1.48e-3 0.00142 31 0.510
#> 11 school.Se… 1.59 1.43 2.80e-4 0.000265 125 0.136
#> 12 school.Se… 1.37 1.05 -2.48e-3 0.000511 118 0.144The following table compares the results of the two approaches.
library(kableExtra)
library(knitr)
tbl_val_gwr_results |>
left_join(tbl_val_results,by="network",suffix = c("",".")) |>
separate_wider_delim(network,delim = ".",names = c("Purpose","Type")) |>
kbl(digits=4) |>
kable_classic_2("hover", full_width = T) |>
add_header_above(c(" " = 2, "GWR" = 6, "Global" = 3)) |>
column_spec(1, bold = T) |>
collapse_rows(columns = 1:2, valign = "top") |>
as_image(width = 7,file = "README_files/figure-gfm/model_table.png")
oldwd <- setwd("../npt/outputdata")
system("gh release download v2023-10-18-08-46-09.827274_commit_57d2dcdd739f665bb8d3fb5540cb6bbd10fcb08f -p combined_network_tile.zip")
unzip("combined_network_tile.zip")
setwd(oldwd)combined_network = st_read(dsn = "../npt/outputdata/combined_network_tile.geojson")
#> Reading layer `combined_network_tile' from data source
#> `C:\Users\ts18jpf\OneDrive - University of Leeds\03_PhD\00_Misc_projects\npt\outputdata\combined_network_tile.geojson'
#> using driver `GeoJSON'
#> Simple feature collection with 563466 features and 26 fields
#> Geometry type: LINESTRING
#> Dimension: XY
#> Bounding box: xmin: -7.54343 ymin: 54.67982 xmax: -0.8615 ymax: 60.76383
#> Geodetic CRS: WGS 84
combined_network
#> Simple feature collection with 563466 features and 26 fields
#> Geometry type: LINESTRING
#> Dimension: XY
#> Bounding box: xmin: -7.54343 ymin: 54.67982 xmax: -0.8615 ymax: 60.76383
#> Geodetic CRS: WGS 84
#> First 10 features:
#> all_fastest_bicycle all_fastest_bicycle_ebike all_fastest_bicycle_go_dutch
#> 1 0 14 0
#> 2 0 0 0
#> 3 0 3 0
#> 4 0 1 0
#> 5 0 0 0
#> 6 0 0 0
#> 7 0 0 0
#> 8 0 0 0
#> 9 0 0 0
#> 10 0 0 0
#> all_quietest_bicycle all_quietest_bicycle_ebike
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> all_quietest_bicycle_go_dutch commute_fastest_bicycle
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> commute_fastest_bicycle_ebike commute_fastest_bicycle_go_dutch
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> commute_quietest_bicycle commute_quietest_bicycle_ebike
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> commute_quietest_bicycle_go_dutch primary_fastest_bicycle
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> primary_fastest_bicycle_ebike primary_fastest_bicycle_go_dutch
#> 1 0 0
#> 2 0 0
#> 3 3 0
#> 4 1 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> primary_quietest_bicycle primary_quietest_bicycle_ebike
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> primary_quietest_bicycle_go_dutch secondary_fastest_bicycle
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> secondary_fastest_bicycle_ebike secondary_fastest_bicycle_go_dutch
#> 1 14 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> secondary_quietest_bicycle secondary_quietest_bicycle_ebike
#> 1 0 0
#> 2 0 0
#> 3 0 0
#> 4 0 0
#> 5 0 0
#> 6 0 0
#> 7 0 0
#> 8 0 0
#> 9 0 0
#> 10 0 0
#> secondary_quietest_bicycle_go_dutch Gradient Quietness
#> 1 0 6 40
#> 2 0 7 40
#> 3 0 0 85
#> 4 0 0 50
#> 5 0 2 70
#> 6 0 6 70
#> 7 0 0 50
#> 8 0 0 70
#> 9 0 6 70
#> 10 0 0 80
#> geometry
#> 1 LINESTRING (-5.03374 54.902...
#> 2 LINESTRING (-5.03983 54.906...
#> 3 LINESTRING (-5.00087 54.904...
#> 4 LINESTRING (-5.00484 54.906...
#> 5 LINESTRING (-5.1116 54.8585...
#> 6 LINESTRING (-5.1116 54.8585...
#> 7 LINESTRING (-4.93467 54.861...
#> 8 LINESTRING (-4.90534 54.869...
#> 9 LINESTRING (-4.89788 54.868...
#> 10 LINESTRING (-4.90534 54.869...tmap_mode("plot")
qtm(combined_network)
### GWR
The following code runs the GWR model for all columns in the
combined_network object
cnet_val_gwr = function(cnet, counts) {
lst_names = names(cnet)[grepl("_", names(cnet))]
# Creating buffer
rnet_buffer20 = cnet |>
st_union() |>
st_buffer(dist = 20)
# Subsetting counts based on buffer
counts_selected = counts[rnet_buffer20,]
# Creating buffer around counts 30 m
counts_buf30 = st_buffer(counts_selected, dist = 30)
# Finding overlapping counts
counts_overlap_30 = st_intersects(counts_buf30, counts_buf30)
# Processing overlaps and aggregating if possible
aggregated_counts =
do.call(rbind,
lapply(unique(counts_overlap_30),
function(x) {
tmp_group = counts_selected[x, ]
# Count aggreagation
simp_data = tmp_group |>
# Removing the direction from the location string
mutate(location = str_remove(location,
"\\s\\w*bound")) |>
st_drop_geometry() |>
# Extracting the first value for the siteID,
# provider and adds up the counts for sites with
# the same 'location'
summarise(across(c("siteID", "provider"),
\(x) head(x, n = 1)),
across(starts_with("count_"), sum),
.by = "location")
simp_group = tmp_group |>
select(siteID) |>
filter(siteID %in% simp_data$siteID)
simp_counts = simp_group |>
left_join(simp_data, by = "siteID") |>
relocate(location, .after = provider) |>
relocate(geometry, .after = count_max)
return(simp_counts)
}))
aggregated_counts$nearest_edge = st_nearest_feature(aggregated_counts,
cnet,
check_crs = T)
val_counts = cbind(aggregated_counts,
st_drop_geometry(cnet)[aggregated_counts$nearest_edge,])
rnet_name = lst_names[1]
val_counts = cbind(aggregated_counts,
st_drop_geometry(cnet)[aggregated_counts$nearest_edge, lst_names]) |> as_Spatial()
reg_rnet =
lapply(lst_names, function(rnet_name) {
form = as.formula(paste(rnet_name, "~ count_mean"))
bw <- bw.gwr(
formula = form,
approach = "AIC",
adaptive = T,
data = val_counts
)
gwr.mod <- gwr.basic(
formula = form,
adaptive = T,
data = val_counts,
bw = bw
)
return(gwr.mod)
})
names(reg_rnet) = lst_names
return(reg_rnet)
}val_results_gwr_last = cnet_val_gwr(cnet = combined_network,
counts = sf_counts)tbl_val_gwr_results_last = tibble(network = names(val_results_gwr_last),
mean_intercept = vapply(val_results_gwr_last,\(x) mean(x$SDF@data[, 1]),FUN.VALUE = 0),
median_intercept = vapply(val_results_gwr_last,\(x) median(x$SDF@data[, 1]),FUN.VALUE = 0),
mean_coef = vapply(val_results_gwr_last,\(x) mean(x$SDF@data[, 2]),FUN.VALUE = 0),
median_coef = vapply(val_results_gwr_last,\(x) median(x$SDF@data[, 2]),FUN.VALUE = 0),
bw = vapply(val_results_gwr_last,\(x) x$GW.arguments$bw,FUN.VALUE = 0),
R2 = vapply(val_results_gwr_last,\(x) x$GW.diagnostic$gw.R2,FUN.VALUE = 0)
)
tbl_val_gwr_results_last |>
arrange(-R2) |>
kbl(digits=4) |>
kable_classic_2("hover", full_width = T) |>
as_image(width = 7,file = "README_files/figure-gfm/val_last_table.png")
The number of trips in the latest build has not been expanded to AADT counts. Using the factors estimated in (here)[https://github.com/nptscot/TT_-Scottish_Household_Survey/commit/2869f2ac845924bf51f3d5262c3f7579f362bdfa], the values will be adjusted.
Also, the number of trips considered for each count will now include the links within a 20 m buffer. It is worth noting that this comparison assumes the intercept as 0.
cnet = st_read(dsn = "../npt/outputdata/combined_network_tile.geojson")
#> Reading layer `combined_network_tile' from data source
#> `C:\Users\ts18jpf\OneDrive - University of Leeds\03_PhD\00_Misc_projects\npt\outputdata\combined_network_tile.geojson'
#> using driver `GeoJSON'
#> Simple feature collection with 563466 features and 26 fields
#> Geometry type: LINESTRING
#> Dimension: XY
#> Bounding box: xmin: -7.54343 ymin: 54.67982 xmax: -0.8615 ymax: 60.76383
#> Geodetic CRS: WGS 84Redefining the function to apply the expansion factor and the buffer for the counts.
library(GWmodel)
cnet_val_gwr_AADT = function(cnet, sf_counts) {
lst_names = names(cnet)[grepl("_", names(cnet))]
# Creating buffer
rnet_buffer30 = cnet |>
st_union() |>
st_buffer(dist = 30)
# Subsetting counts based on buffer
counts_selected = sf_counts[rnet_buffer30,]
# Creating buffer around counts 30 m
counts_buf30 = st_buffer(counts_selected, dist = 30)
# Finding overlapping counts
counts_overlap_30 = st_intersects(counts_buf30, counts_buf30)
# Processing overlaps and aggregating if possible
aggregated_counts =
do.call(rbind,
lapply(unique(counts_overlap_30),
function(x) {
tmp_group = counts_selected[x, ]
# Count aggreagation
simp_data = tmp_group |>
# Removing the direction from the location string
mutate(location = str_remove(location,
"\\s\\w*bound")) |>
st_drop_geometry() |>
# Extracting the first value for the siteID,
# provider and adds up the counts for sites with
# the same 'location'
summarise(across(c("siteID", "provider"),
\(x) head(x, n = 1)),
across(starts_with("count_"), sum),
.by = "location")
simp_group = tmp_group |>
select(siteID) |>
filter(siteID %in% simp_data$siteID)
simp_counts = simp_group |>
left_join(simp_data, by = "siteID") |>
relocate(location, .after = provider) |>
relocate(geometry, .after = count_max)
return(simp_counts)
}))
# Buffer for the aggregated counts
aggr_countrs_buffer = st_buffer(aggregated_counts, dist = 20)
# Subsetting the network based on the buffer
cnet_selected = cnet[aggr_countrs_buffer,]
# Spatial join to detect the network links within the buffer
Agg_network_siteID = cnet_selected |>
st_join(aggr_countrs_buffer) |>
st_drop_geometry() |>
select(all_fastest_bicycle:secondary_quietest_bicycle_go_dutch,
siteID) |>
summarise(across(contains("bicycle"), sum), .by = siteID)
# Join with the aggregated counts and factoring for AADT
val_counts = aggregated_counts |>
left_join(Agg_network_siteID, by = "siteID") |>
# Hard-coded AADT factors
mutate(across(contains("commute"), \(x) x * 1.7373),
across(contains("ary"), \(x) x * 1.5947)) |>
mutate(across(
contains('all'),
~ get(str_replace(cur_column(), "^all", "commute")) +
get(str_replace(cur_column(), "^all", "primary")) +
get(str_replace(cur_column(), "^all", "secondary"))
)) |>
drop_na(all_fastest_bicycle:secondary_quietest_bicycle_go_dutch) |>
as_Spatial()
reg_rnet =
lapply(lst_names, function(rnet_name) {
form = as.formula(paste(rnet_name, "~ count_mean + 0"))
bw <- bw.gwr(
formula = form,
approach = "AIC",
adaptive = T,
data = val_counts
)
gwr.mod <- gwr.basic(
formula = form,
adaptive = T,
data = val_counts,
bw = bw
)
return(gwr.mod)
})
names(reg_rnet) = lst_names
return(reg_rnet)
}set.seed(123456)
val_results_gwr_last_AADT = cnet_val_gwr_AADT(cnet,
sf_counts)
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4852.424
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4796.848
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4783.099
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 4786.686
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 4781.444
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 4781.341
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 4786.376
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4781.801
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 4782.189
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4781.934
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4781.344
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4781.934
#> Adaptive bandwidth (number of nearest neighbours): 115 AICc value: 4781.811
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4781.934
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4781.934
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4781.721
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4781.721
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4781.801
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4781.801
#> Adaptive bandwidth (number of nearest neighbours): 111 AICc value: 4781.676
#> Adaptive bandwidth (number of nearest neighbours): 111 AICc value: 4781.676
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 4781.853
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 4781.853
#> Adaptive bandwidth (number of nearest neighbours): 109 AICc value: 4779.952
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4781.344
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4781.344
#> Adaptive bandwidth (number of nearest neighbours): 115 AICc value: 4781.811
#> Adaptive bandwidth (number of nearest neighbours): 115 AICc value: 4781.811
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4781.934
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4781.934
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4781.721
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4781.721
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6035.532
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 6002.394
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5989.244
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 5992.276
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 5988.775
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 5991.633
#> Adaptive bandwidth (number of nearest neighbours): 100 AICc value: 5988.301
#> Adaptive bandwidth (number of nearest neighbours): 95 AICc value: 5988.882
#> Adaptive bandwidth (number of nearest neighbours): 101 AICc value: 5988.568
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5988.849
#> Adaptive bandwidth (number of nearest neighbours): 99 AICc value: 5988.636
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5988.849
#> Adaptive bandwidth (number of nearest neighbours): 98 AICc value: 5988.83
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5988.849
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5988.849
#> Adaptive bandwidth (number of nearest neighbours): 96 AICc value: 5988.812
#> Adaptive bandwidth (number of nearest neighbours): 96 AICc value: 5988.812
#> Adaptive bandwidth (number of nearest neighbours): 95 AICc value: 5988.882
#> Adaptive bandwidth (number of nearest neighbours): 95 AICc value: 5988.882
#> Adaptive bandwidth (number of nearest neighbours): 94 AICc value: 5988.735
#> Adaptive bandwidth (number of nearest neighbours): 94 AICc value: 5988.735
#> Adaptive bandwidth (number of nearest neighbours): 93 AICc value: 5988.775
#> Adaptive bandwidth (number of nearest neighbours): 93 AICc value: 5988.775
#> Adaptive bandwidth (number of nearest neighbours): 92 AICc value: 5988.997
#> Adaptive bandwidth (number of nearest neighbours): 92 AICc value: 5988.997
#> Adaptive bandwidth (number of nearest neighbours): 91 AICc value: 5989.051
#> Adaptive bandwidth (number of nearest neighbours): 91 AICc value: 5989.051
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5989.244
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5989.244
#> Adaptive bandwidth (number of nearest neighbours): 89 AICc value: 5989.345
#> Adaptive bandwidth (number of nearest neighbours): 89 AICc value: 5989.345
#> Adaptive bandwidth (number of nearest neighbours): 88 AICc value: 5989.411
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6026.99
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 5991.422
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5980.976
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 5983.81
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 5980.047
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 5980.971
#> Adaptive bandwidth (number of nearest neighbours): 100 AICc value: 5980.202
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 5980.786
#> Adaptive bandwidth (number of nearest neighbours): 103 AICc value: 5980.177
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 5979.985
#> Adaptive bandwidth (number of nearest neighbours): 108 AICc value: 5980.241
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 5980.047
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 5979.985
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4981.513
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4932.045
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4921.52
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 4924.063
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 4920.499
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 4919.515
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 4923.613
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4920.752
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 4919.22
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 4922.225
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 4919.22
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6397.743
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 6378.091
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 6373.633
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 6375.175
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 6373.154
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 6373.343
#> Adaptive bandwidth (number of nearest neighbours): 100 AICc value: 6373.29
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 6374.034
#> Adaptive bandwidth (number of nearest neighbours): 103 AICc value: 6373.389
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 6373.167
#> Adaptive bandwidth (number of nearest neighbours): 104 AICc value: 6373.06
#> Adaptive bandwidth (number of nearest neighbours): 104 AICc value: 6373.06
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6271.953
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 6252.942
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 6249.74
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 6251.139
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 6249.113
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 6248.672
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 6249.959
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 6249.849
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 6248.635
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 6249.567
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 6248.635
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4845.904
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4790.102
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4776.275
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 4779.93
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 4774.572
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 4774.492
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 4779.567
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4774.922
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 4775.34
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4775.058
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4774.494
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4775.058
#> Adaptive bandwidth (number of nearest neighbours): 115 AICc value: 4774.934
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4775.058
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4775.058
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4774.838
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4774.838
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4774.922
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4774.922
#> Adaptive bandwidth (number of nearest neighbours): 111 AICc value: 4774.797
#> Adaptive bandwidth (number of nearest neighbours): 111 AICc value: 4774.797
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 4774.981
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 4774.981
#> Adaptive bandwidth (number of nearest neighbours): 109 AICc value: 4773.067
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4774.494
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4774.494
#> Adaptive bandwidth (number of nearest neighbours): 115 AICc value: 4774.934
#> Adaptive bandwidth (number of nearest neighbours): 115 AICc value: 4774.934
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4775.058
#> Adaptive bandwidth (number of nearest neighbours): 114 AICc value: 4775.058
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4774.838
#> Adaptive bandwidth (number of nearest neighbours): 113 AICc value: 4774.838
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6011.584
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 5977.9
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5964.45
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 5967.474
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 5963.943
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 5966.86
#> Adaptive bandwidth (number of nearest neighbours): 100 AICc value: 5963.469
#> Adaptive bandwidth (number of nearest neighbours): 95 AICc value: 5964.055
#> Adaptive bandwidth (number of nearest neighbours): 101 AICc value: 5963.745
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5964.011
#> Adaptive bandwidth (number of nearest neighbours): 99 AICc value: 5963.796
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5964.011
#> Adaptive bandwidth (number of nearest neighbours): 98 AICc value: 5963.997
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5964.011
#> Adaptive bandwidth (number of nearest neighbours): 97 AICc value: 5964.011
#> Adaptive bandwidth (number of nearest neighbours): 96 AICc value: 5963.98
#> Adaptive bandwidth (number of nearest neighbours): 96 AICc value: 5963.98
#> Adaptive bandwidth (number of nearest neighbours): 95 AICc value: 5964.055
#> Adaptive bandwidth (number of nearest neighbours): 95 AICc value: 5964.055
#> Adaptive bandwidth (number of nearest neighbours): 94 AICc value: 5963.941
#> Adaptive bandwidth (number of nearest neighbours): 94 AICc value: 5963.941
#> Adaptive bandwidth (number of nearest neighbours): 93 AICc value: 5963.984
#> Adaptive bandwidth (number of nearest neighbours): 93 AICc value: 5963.984
#> Adaptive bandwidth (number of nearest neighbours): 92 AICc value: 5964.203
#> Adaptive bandwidth (number of nearest neighbours): 92 AICc value: 5964.203
#> Adaptive bandwidth (number of nearest neighbours): 91 AICc value: 5964.257
#> Adaptive bandwidth (number of nearest neighbours): 91 AICc value: 5964.257
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5964.45
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5964.45
#> Adaptive bandwidth (number of nearest neighbours): 89 AICc value: 5964.552
#> Adaptive bandwidth (number of nearest neighbours): 89 AICc value: 5964.552
#> Adaptive bandwidth (number of nearest neighbours): 88 AICc value: 5964.619
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 5977.602
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 5940.603
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 5929.617
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 5932.616
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 5928.647
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 5929.613
#> Adaptive bandwidth (number of nearest neighbours): 100 AICc value: 5928.806
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 5929.444
#> Adaptive bandwidth (number of nearest neighbours): 103 AICc value: 5928.797
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 5928.571
#> Adaptive bandwidth (number of nearest neighbours): 108 AICc value: 5928.828
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 5928.647
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 5928.571
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4973.578
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4923.486
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4912.77
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 4915.341
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 4911.711
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 4910.746
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 4914.917
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 4911.957
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 4910.443
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 4913.502
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 4910.443
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6365.481
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 6344.848
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 6340.032
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 6341.855
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 6339.52
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 6339.784
#> Adaptive bandwidth (number of nearest neighbours): 100 AICc value: 6339.63
#> Adaptive bandwidth (number of nearest neighbours): 110 AICc value: 6340.447
#> Adaptive bandwidth (number of nearest neighbours): 103 AICc value: 6339.748
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 6339.526
#> Adaptive bandwidth (number of nearest neighbours): 104 AICc value: 6339.433
#> Adaptive bandwidth (number of nearest neighbours): 104 AICc value: 6339.433
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 6198.514
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 6177.156
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 6173.168
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 6175.143
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 6172.486
#> Adaptive bandwidth (number of nearest neighbours): 117 AICc value: 6172.168
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 6173.695
#> Adaptive bandwidth (number of nearest neighbours): 112 AICc value: 6173.3
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 6172.112
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 6173.23
#> Adaptive bandwidth (number of nearest neighbours): 118 AICc value: 6172.112
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 2296.459
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2293.902
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 2296.115
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 2295.782
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 2293.96
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 2295.217
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 2293.814
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 2293.826
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 2293.685
#> Adaptive bandwidth (number of nearest neighbours): 130 AICc value: 2293.487
#> Adaptive bandwidth (number of nearest neighbours): 131 AICc value: 2293.456
#> Adaptive bandwidth (number of nearest neighbours): 131 AICc value: 2293.456
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 3662.399
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 3659.476
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 3660.303
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 3661.243
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 3659.524
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 3660.349
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 3659.398
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 3659.393
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 3659.295
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 3659.295
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4049.339
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4046.222
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4048.339
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4047.83
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4046.379
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 4047.028
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4046.114
#> Adaptive bandwidth (number of nearest neighbours): 123 AICc value: 4046.123
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 4046.113
#> Adaptive bandwidth (number of nearest neighbours): 130 AICc value: 4046.01
#> Adaptive bandwidth (number of nearest neighbours): 131 AICc value: 4046.015
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 4046.113
#> Adaptive bandwidth (number of nearest neighbours): 130 AICc value: 4046.01
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 2664.633
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2664.06
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 2666.872
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 2665.206
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 2665.088
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 2664.918
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 2664.48
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 2664.128
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 2664.314
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 2663.981
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 2664.128
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2664.06
#> Adaptive bandwidth (number of nearest neighbours): 135 AICc value: 2664.098
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2664.06
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2664.06
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 2663.981
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4074.767
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4074.159
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4076.886
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4075.17
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4075.047
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 4074.848
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4074.477
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4074.187
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 4074.35
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 4074.107
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4074.187
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4074.159
#> Adaptive bandwidth (number of nearest neighbours): 135 AICc value: 4074.198
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4074.159
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4074.159
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 4074.107
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4419.949
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4419.351
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4422.194
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4420.352
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4420.238
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 4420.041
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4419.669
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4419.379
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 4419.542
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 4419.3
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4419.379
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4419.351
#> Adaptive bandwidth (number of nearest neighbours): 135 AICc value: 4419.391
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4419.351
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4419.351
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 4419.3
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 2249.094
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2236.913
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 2235.899
#> Adaptive bandwidth (number of nearest neighbours): 63 AICc value: 2238.188
#> Adaptive bandwidth (number of nearest neighbours): 106 AICc value: 2236.413
#> Adaptive bandwidth (number of nearest neighbours): 79 AICc value: 2235.801
#> Adaptive bandwidth (number of nearest neighbours): 73 AICc value: 2236.651
#> Adaptive bandwidth (number of nearest neighbours): 83 AICc value: 2235.538
#> Adaptive bandwidth (number of nearest neighbours): 85 AICc value: 2235.455
#> Adaptive bandwidth (number of nearest neighbours): 87 AICc value: 2235.947
#> Adaptive bandwidth (number of nearest neighbours): 84 AICc value: 2235.477
#> Adaptive bandwidth (number of nearest neighbours): 85 AICc value: 2235.455
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4128.019
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4122.28
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4123.963
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4124.815
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4122.034
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 4123.004
#> Adaptive bandwidth (number of nearest neighbours): 124 AICc value: 4121.835
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4121.979
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 4121.599
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4121.858
#> Adaptive bandwidth (number of nearest neighbours): 122 AICc value: 4121.727
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4121.858
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 4121.857
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4121.858
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4121.858
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 4121.599
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4547.675
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4541.947
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4544.006
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4544.084
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4541.611
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 4542.887
#> Adaptive bandwidth (number of nearest neighbours): 124 AICc value: 4541.391
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4541.561
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 4541.142
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4541.478
#> Adaptive bandwidth (number of nearest neighbours): 122 AICc value: 4541.285
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4541.478
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 4541.451
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4541.478
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 4541.478
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 4541.142
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 2382.942
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 2370.887
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 2371.408
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 2375.737
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 2369.46
#> Adaptive bandwidth (number of nearest neighbours): 107 AICc value: 2370.904
#> Adaptive bandwidth (number of nearest neighbours): 124 AICc value: 2369.363
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 2369.796
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 2368.528
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 2369.21
#> Adaptive bandwidth (number of nearest neighbours): 122 AICc value: 2368.964
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 2369.21
#> Adaptive bandwidth (number of nearest neighbours): 121 AICc value: 2369.165
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 2369.21
#> Adaptive bandwidth (number of nearest neighbours): 120 AICc value: 2369.21
#> Adaptive bandwidth (number of nearest neighbours): 119 AICc value: 2368.528
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4577.192
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4575.486
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4578.219
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4576.91
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4576.469
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 4576.313
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4575.712
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4575.493
#> Adaptive bandwidth (number of nearest neighbours): 129 AICc value: 4575.68
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 4575.468
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4575.493
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4575.486
#> Adaptive bandwidth (number of nearest neighbours): 135 AICc value: 4575.51
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4575.486
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4575.486
#> Adaptive bandwidth (number of nearest neighbours): 133 AICc value: 4575.468
#> Adaptive bandwidth (number of nearest neighbours): 204 AICc value: 4946.776
#> Adaptive bandwidth (number of nearest neighbours): 134 AICc value: 4945.025
#> Adaptive bandwidth (number of nearest neighbours): 90 AICc value: 4947.969
#> Adaptive bandwidth (number of nearest neighbours): 160 AICc value: 4946.456
#> Adaptive bandwidth (number of nearest neighbours): 116 AICc value: 4946.021
#> Adaptive bandwidth (number of nearest neighbours): 143 AICc value: 4945.86
#> Adaptive bandwidth (number of nearest neighbours): 126 AICc value: 4945.256
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4945.024
#> Adaptive bandwidth (number of nearest neighbours): 140 AICc value: 4945.495
#> Adaptive bandwidth (number of nearest neighbours): 136 AICc value: 4945.024library(kableExtra)
tbl_val_gwr_results_last_AADT = tibble(
network = names(val_results_gwr_last_AADT),
mean_coef = vapply(val_results_gwr_last_AADT,
\(x) mean(x$SDF@data[, 1]),FUN.VALUE = 0),
median_coef = vapply(val_results_gwr_last_AADT,
\(x) median(x$SDF@data[, 1]),FUN.VALUE = 0),
bw = vapply(val_results_gwr_last_AADT,
\(x) x$GW.arguments$bw,FUN.VALUE = 0),
R2 = vapply(val_results_gwr_last_AADT,
\(x) x$GW.diagnostic$gw.R2,FUN.VALUE = 0)
)
tbl_val_gwr_results_last_AADT |>
arrange(-R2) |>
kbl(digits=3) |>
kable_classic_2("hover", full_width = T) |>
as_image(width = 10,file = "README_files/figure-gfm/val_last_table_AADT.png")
A closer look at the distribution of the coefficients count_mean for
the all_fastest_bicycleshows that bike flows might be overestimated in
Edinburgh by the model, while in Glasgow the estimates are lower than
observed.
best_network = val_results_gwr_last_AADT$all_fastest_bicycle$SDF |> st_as_sf()
tmap_mode("plot")
mymap = tm_shape(best_network)+
tm_dots(col = "count_mean",palette = "RdYlGn",
midpoint=1,size = 0.3,alpha=0.4)+
tm_basemap()+
tm_layout(legend.position = c("left", "top"))
tmap_save(mymap,filename = "README_files/figure-gfm/coef_maps.png",units = "cm",width = 6)
#> | | | 0% | | | 1% | |. | 2% [unnamed-chunk-132] | |.. | 3% [unnamed-chunk-133] | |.. | 4% | |.. | 5% [unnamed-chunk-134] | |... | 5% | |... | 6% [unnamed-chunk-135] | |... | 7% | |.... | 8% [unnamed-chunk-136] | |..... | 9% [unnamed-chunk-137] | |..... | 10% | |..... | 11% [unnamed-chunk-138] | |...... | 11% | |...... | 12% [unnamed-chunk-139] | |...... | 13% | |....... | 14% [unnamed-chunk-140] | |....... | 15% | |........ | 15% [unnamed-chunk-141] | |........ | 16% | |........ | 17% [unnamed-chunk-142] | |......... | 18% [unnamed-chunk-143] | |.......... | 19% | |.......... | 20% [unnamed-chunk-144] | |.......... | 21% | |........... | 21% [unnamed-chunk-145] | |........... | 22% | |........... | 23% [unnamed-chunk-146] | |............ | 24% [unnamed-chunk-147] | |............. | 25% | |............. | 26% [unnamed-chunk-148] | |............. | 27% | |.............. | 27% [unnamed-chunk-149] | |.............. | 28% | |............... | 29% [unnamed-chunk-150] | |............... | 30% | |............... | 31% [unnamed-chunk-151] | |................ | 31% | |................ | 32% [unnamed-chunk-152] | |................ | 33% | |................. | 34% [unnamed-chunk-153] | |.................. | 35% [unnamed-chunk-154] | |.................. | 36% | |.................. | 37% [unnamed-chunk-155] | |................... | 37% | |................... | 38% [unnamed-chunk-156] | |................... | 39% | |.................... | 40% [unnamed-chunk-157] | |..................... | 41% [unnamed-chunk-158] | |..................... | 42% | |..................... | 43% [unnamed-chunk-159] | |...................... | 44% [unnamed-chunk-160] | |....................... | 45% | |....................... | 46% [unnamed-chunk-161] | |....................... | 47% | |........................ | 47% [unnamed-chunk-162] | |........................ | 48% | |........................ | 49% [unnamed-chunk-163] | |......................... | 50% [unnamed-chunk-164] | |.......................... | 51% | |.......................... | 52% [unnamed-chunk-165] | |.......................... | 53% | |........................... | 53% [unnamed-chunk-166] | |........................... | 54% | |........................... | 55% [unnamed-chunk-167] | |............................ | 56% [unnamed-chunk-168] | |............................. | 57% | |............................. | 58% [unnamed-chunk-169] | |............................. | 59% | |.............................. | 60% [unnamed-chunk-170] | |............................... | 61% [unnamed-chunk-171] | |............................... | 62% | |............................... | 63% [unnamed-chunk-172] | |................................ | 63% | |................................ | 64% [unnamed-chunk-173] | |................................ | 65% | |................................. | 66% [unnamed-chunk-174] | |.................................. | 67% [unnamed-chunk-175] | |.................................. | 68% | |.................................. | 69% [unnamed-chunk-176] | |................................... | 69% | |................................... | 70% [unnamed-chunk-177] | |................................... | 71% | |.................................... | 72% [unnamed-chunk-178] | |.................................... | 73% | |..................................... | 73% [unnamed-chunk-179] | |..................................... | 74% | |..................................... | 75% [unnamed-chunk-180] | |...................................... | 76% [unnamed-chunk-181] | |....................................... | 77% | |....................................... | 78% [unnamed-chunk-182] | |....................................... | 79% | |........................................ | 79% [unnamed-chunk-183] | |........................................ | 80% | |........................................ | 81% [unnamed-chunk-184] | |......................................... | 82% [unnamed-chunk-185] | |.......................................... | 83% | |.......................................... | 84% [unnamed-chunk-186] | |.......................................... | 85% | |........................................... | 85% [unnamed-chunk-187] | |........................................... | 86% | |............................................ | 87% [unnamed-chunk-188] | |............................................ | 88% | |............................................ | 89% [unnamed-chunk-189] | |............................................. | 89% | |............................................. | 90% [unnamed-chunk-190] | |............................................. | 91% | |.............................................. | 92% [unnamed-chunk-191] | |............................................... | 93% [unnamed-chunk-192] | |............................................... | 94% | |............................................... | 95% [unnamed-chunk-193] | |................................................ | 95% | |................................................ | 96% [unnamed-chunk-194] | |................................................ | 97% | |................................................. | 98% [unnamed-chunk-195] [unnamed-chunk-196] | |..................................................| 99% | |..................................................| 100% [unnamed-chunk-197]
#> [1] "counts.R"
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent