personlin / gmpewebapp Goto Github PK

AUTHOR: Robert Clements
DATE: March 7, 2013
SUMMARY: Readme.txt file for GMPE evaluation R "shiny" web application
----------------------------------------------------------------------

Required packages to run app:

ggplot2
verification
lme4
grid
plyr
mgcv
RColorBrewer
ggmap

Use install.packages("package.name") to install these.

------------------------------------------------------
------------------------------------------------------


To run the app:

If all files are on your machine already, then do this:

setwd("path to directory that holds the ui.R, server.R, jp, and sample.csv files")
library(shiny)
runApp()

Or, if you don't have the files on your machine, you can do this:

library(shiny)
shiny::runGitHub('gmpeWebapp', 'r-clements')

------------------------------------------------------
------------------------------------------------------

To use the app:

-First choose from the dropdown menu if you want to do "Residual analysis" or look at "Scores"
-A sample dataset is loaded automatically. If you have your own dataset that you would like to default to, just name it "sample.csv".
-You can also choose a different file to use by clicking the checkbox "Choose a csv file ..."

Most everything else is pretty self explanatory. The only thing that might cause confusion is the linear and mixed models. Currently, I have implemented both a linear model fit to the residuals, as well as a mixed effects model fit to the residuals. The mixed effects model automatically includes a random effect for "event.num".

Lastly, regarding the two spatial residuals plots, the "smooth" option fits a thin-plate spline to the residuals using the generalized cross-validation criteria in the gam function. This plot is just to get a general overview of the spatial residuals, but I may provide more control on this smoothing to the user if it's useful.