ellisp / ellisp.github.io Goto Github PK

unnecessarily high resolution

See http://ellisp.github.io/blog/2016/04/09/nzelect3/

where did they go?

Dear Peter,

On your about page you mention StatsBlogs. The URL statsblogs.com seems to be broken. By any chance, do you know if they moved their blog to another URL or if they simply stopped their activity?

I'd be interested to read their blog if it is still available.

Thanks in advance.
Best,
Antoine

Hi Peter,
I was stepping through your code on the blog below
http://ellisp.github.io/blog/2015/10/10/X13ARIMA-SEATS.html
I am unable to download the below dataset..do you by any chance happen to have the tmp.rda file stored?

download.file("https://github.com/ellisp/ellisp.github.io/blob/master/data/Electronic card transactions by industry group Monthly.rda?raw=true",
mode = "wb", # ie binary
destfile = "tmp.rda")

many thanks for all your help
heena

Hi Peter,

Love your blog. Your analysis, shinys, datavis and writing are always great, i read each post religiously. I've been modelling demography for around 18 months so I've enjoyed your recent ventures into this space.

I've got a topic that I'd love to see a blog post on, a topic that i don't think has ever been blogged on (at least not that i can find). Most cohort component demographic modelling exists in an unconstrained environment, i.e. without 'carrying capacity' constraints, such as available land for housing.

So the potential blog topic has two elements:
a) How would you go about the spatio-temporal modelling of carry capacity constraints using census data (such that you can forecast sub regions)? I mention census data, but i'd be interested in how else you model this too, perhaps aerial imagery?
b) How would you go about incorporating the carrying capacity modelling into the cohort component model? How would you back-propagate the changes due to carrying capacity such that the cohort component model accounting still adds up?

I've recently done the above modelling. But you are in a different league to me, so i'd be fascinated to hear how you might go about it. For what it is worth I used a spatio-temporal GAM on population densities by small statistical regions to estimate regions of change. I transformed this into a time series index for each region - the index only changed the cohort component model in regions that have been growing very rapidly for a long period of time. Where the index changed the cohort component model i used net migration to soak up these changes such that the cohort component model accounting added up.

Sorry, that 'issue' got a little long. Eager to hear your thoughts.

Alex

beta versus quasibinomial regression

http://shootingtracker.com/wiki/Mass_Shootings_in_2015

IT things statisticians need to know - data modelling, dev/test/prod, what a BA does, how project gates work, difference between a solution and enterprise architect

confidence bands for density (using bootstrap, as opposed to this method http://www.r-bloggers.com/delta-method-confidence-bands-for-gaussian-mixture-density-can-behave-badly/ )

putting shiny apps in iframes for R-bloggers (noticed lots of people don't)

streamcharts

sankey charts

fonts

Static treemaps

Interactive v static treemaps

force network plots

Jekyll and Ruby

The whole toolchain

Analysts need to use version control

R doesn't matter, code-based matters

Automating commentary in MTAGDP
The MTAGDP creation process
The MTAGDP app

The Regional Economic Activity Report database

The overall Regional Economic Activity Report toolchain

Seasonally adjusting NZ economic data

Census 2013 - reusing meshblocks

NZ maps

{mbie}

{mbiemaps}

elections in NZ

The Regional Tourism Indicators

ELO ratings simulations

Hi Peter,
I am putting together a conference presentation and want to use some of your code in it (e.g. the code from https://ellisp.github.io/blog/2015/08/15/importing-nzis-surf). Your blog material is licensed as CC-BY-SA. Do you have a preferred form for attribution?
Cheers
Ross

Hi

Paul Murrell here from the Department of Statistics at The University of Auckland ([email protected]). I am just trying to get in touch to ask about how (whether) you get on with producing macron-accented characters in R plots. Sorry for posting here, but I could not find an email and I (stubbornly) refuse to join walled gardens like Facebook and LinkedIn just to get in touch with someone. It's just my small act of resistance :)

Hope you get this and are happy to get in touch.

Paul

when to use twitter, email, linked in. facebook for personal friends.

Hi, The 'overall outcomes' chart for Model A seems to have gone missing in the last couple of days. The equivalent on Model B is still there. Thanks for the good work! I suspect that the predictive capabilities of the models will be being severely challenged by the rapid rates of change in voter intentions. Cheers, Derek

needs to match the overall page

Is it worth doing an in-depth look at Victoria’s new cases of Covid-19 data between the dates of 9 July (Initial Melbourne lockdown), 22 July (mandatory mask wearing in Melbourne), and 3 August (introduction of Stage 4 restrictions). All recommendations to introduce mandatory mask wearing has been based on advice, not proof, that they help control spread of the virus. It appears there is a very short period of data that shows mask wearing is driving the current decrease in new cases.

Respected Sir,
I am Dr. Asif and I am from Bangladesh. I am an epidemiology student and learning infectious disease modelling with R. I am currently working as surveillance officer in Communicable disease Control ,DGHS . We are trying to write a paper on Covid-19 fatalities according to age and gender. But due to very restricted data source I am having some problem. After reading your blog I believe I really need your Suggestion and help. I am trying to contact with you through your e-mail but cant get any, and I have also tried to communicate with you in linked in. I really need your valuable help.

Sincerely
Dr. K.M. Rukhsad Asif Zaman

"Clerical and Adminsitrative Workers"

In the paragraph discussing LinkedIn profiles, you accidental wrote “CE” instead of “CEO.” I only brought this up because you later made fun of their typo.

https://support.google.com/webmasters/answer/183669

Since February 2017 images resize properly and look better in mobile devices. Go through old posts and fix this too (means using html tags for img src links).

Hi Peter, I'm not sure if this is the best place to do this, but I just wanted to touch base and say hi.

I'm a reporter with Fairfax and I'm going to be doing a lot of reporting on polls, voting demographics, geography etc in the coming weeks and have found your blog to be a brilliant resource.

I'm currently delving into the NZES and am hoping to do some reporting on that leading up to the election. I code in Python mostly, so sadly I may not be able to make best use of you Relect package - perhaps I will have to upskill and learn some more R in time for 2020.

Anyway I just wanted to get in touch given we seem to be interested in similar datasets and to say thanks for the excellent work.

Cheers,
Andy Fyers

Dear Peter,

On your post: Fixing scientific publishing and peer review, you refer to a retraction in Nature on the mental health impact of gaming in Africa (second paragraph). However, the retraction was in the journal Scientific Reports. Please could you correct the blog post accordingly?

Kind Regards
Michael

Hi Peter,
We have had some complaints that h2o::as.h2o() is very slow for sparse matrices and so we came across your awesome blog post on the topic. We are planning to implement your solution (unless you wanted to submit a PR directly).

Thanks!
-Erin