search

kasaai / gisadata

GNU Affero General Public License v3.0
2 stars 2 forks source link

Reconciliation with the PDF #38

Open ResearchActuary opened 3 years ago

ResearchActuary commented 3 years ago

Not sure who else has looked at the exhibits in the PDFs, but I can’t match them to the data in the .CSV files. I’m trying something very simple like checking the number of claims in Alberta TPL, first half of 2004, first evaluation lag. Page 26 (shown as “Page 22”) of the file “AUTO7001-AB_2018.PDF” shows 8,065 claims. Looking at the data imported from the file “AUTO7001-AB-201812-20190312150247.CSV”, I can’t sort out any set of data which would generate that number.

The code to extract the data from the .zips is not concise enough for the issue, but happy to post it, if that helps.

carpenn commented 3 years ago

I've been reviewing and updating findings on the slack thread. All figures reconcile though the appropriate filtering and aggregating needed to get to the pdf figures is at times involved. There is one small issue with the triangle routine for the atlantics aggregation which I've flagged with Marco.

carpenn commented 3 years ago

I've tried to create a pull request for the fixes to generate_triangle_data.R but am have git authentication issues from Rstudio. Hence adding corrected code as a zip file to this post int the hope that others with a working Git connection can make the changes.

generate_triangle_data.zip

carpenn commented 3 years ago

OK have eventually found a way of getting local git to push to here and can see fix has been applied. (Apologies though as I meant create a pull request for others to approve)