Read Survival Analysis materials

[bujowskis]

• [x] Read the papers
• [x] Make notes
• [x] Meet up to clarify concerns

[lukik45]

@white07S should we just familiarise ourselves with these two links, or sth more? https://sphweb.bumc.bu.edu/otlt/MPH-Modules/BS/BS704_Survival/BS704_Survival6.html https://square.github.io/pysurvival/

[white07S]

@white07S should we just familiarise ourselves with these two links, or sth more? https://sphweb.bumc.bu.edu/otlt/MPH-Modules/BS/BS704_Survival/BS704_Survival6.html https://square.github.io/pysurvival/

Understanding the underlying ideas presented in this two links are enough, there are some different models like COX, I would suggest to get an overview of this model.

[white07S]

• [x] Only understand the underlying idea of the COX model

[bujowskis]

So I went through the materials and there are two concerns I have to raise, both regarding this article

(...) There are several important assumptions for the appropriate use of Cox proportional hazards regression model

Independence of survival times between distinct individuals in the sample

In our case, are the stocks independent for the appropriate use of Cox? Still unsure whether to think of distinct individuals as multiple stocks or some points within a given stock, but in both cases think of the following:

• stock on stock - impact analysis and correlations
• within a stock - trends, sentiment analysis, etc.

Constant hazard ratio over time

Is it going to be constant and fulfil that assumption?

[white07S]

Yes it is enough but considering COX have its own assumptions, we need to derive our own assumptions while modelling our model

[bujowskis]

is it going to affect anything we have to do before paper proof and model? The coding parts? @white07S? Otherwise, it'd be more appropriate to start a new Todo issue and make deriving our assumptions a checkbox in it, close this one

[white07S]

Yeah, that's a great idea.