Updating life section

[Pat-Reen]

Adding in subsections on decision trees and bayesian applications.

[JackyP]

Hi @Pat-Reen,

Thank you for the contribution.

Some minor issues picking up from review:

• R: Life Industry Stats in # Tableau and R has an extra "#"
• The new articles just say "Background" and "Overview" on the sidebar, can you add a more descriptive title?
• Libraries section in Bayes doesn't display the libraries
• "Good interpretability and Robustness," has Robustness randomly capitalised
• [Data description](Framingham Longitudinal Data Documentation.pdf) isn't showing up as a link. Is the pdf copyright available?
• In the GBDT vs XGBoost section, I'd argue that XGBoost is a kind of GBDT, one that brings innovations to the algorithm. The docs describe themselves that way too - "XGBoost provides a parallel tree boosting (also known as GBDT, GBM)". Perhaps it could be rephrased a little e.g. "Original GBDT vs XGBoost"?

Another general q's I had: How are you getting NN's to do well on sparse data vs GBDT?

I also just want to say it's really cool to see MCMC and R XGBoost on the cookbook.

[jzhc99]

Thanks Jacky for review. I have made some changes to the Tree model section and will rephrase the description. Agree that xgboost is an improved algorithm of GBDT.

I haven't spent too much time on training life data using NN and the preliminary result was worse than tree models. I would expect NN requires dense data and large dataset to work well. I will speak to Pat about this and see if we can add something.

[Pat-Reen]

Replaced by pull request #36. Addresses the feedback from @JackyP.

[Pat-Reen]

I have updated. Replaced with PR #36.

On Tue, 15 Mar 2022, 11:41 am Patrick Reen, @.***> wrote:

Thanks both. I owe you some edits as well. Will get to in soon. Thanks!

On Tue, 15 Mar 2022, 9:03 am Charlie, @.***> wrote:

Thanks Jackie for review. I have made some changes to the Tree model section and will rephrase the description. Agree that xgboost is an improved algorithm of GBDT.

I haven't spent too much time on training life data using NN and the preliminary result was worse than tree models. I would expect NN requires dense data and large dataset to work well. I will speak to Pat about this and see if we can add something.

— Reply to this email directly, view it on GitHub https://github.com/ActuariesInstitute/cookbook/pull/32#issuecomment-1067346731, or unsubscribe https://github.com/notifications/unsubscribe-auth/AV2VKIH4XGGCC2QYPOZXNCDU76ZUTANCNFSM5NQVDHKA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>

[Pat-Reen]

Thanks both. I owe you some edits as well. Will get to in soon. Thanks!

On Tue, 15 Mar 2022, 9:03 am Charlie, @.***> wrote:

Thanks Jackie for review. I have made some changes to the Tree model section and will rephrase the description. Agree that xgboost is an improved algorithm of GBDT.

I haven't spent too much time on training life data using NN and the preliminary result was worse than tree models. I would expect NN requires dense data and large dataset to work well. I will speak to Pat about this and see if we can add something.

— Reply to this email directly, view it on GitHub https://github.com/ActuariesInstitute/cookbook/pull/32#issuecomment-1067346731, or unsubscribe https://github.com/notifications/unsubscribe-auth/AV2VKIH4XGGCC2QYPOZXNCDU76ZUTANCNFSM5NQVDHKA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you were mentioned.Message ID: @.***>