Lightning Talk - Survival Analysis

[thisisashukla]

• Abstract (2-3 lines)

  Survival analysis was initially used by medical practitioners and actuaries to deal with problems such as:

  • What is life expectancy of patients in cohort A as compared to cohort B (where one of the cohorts would be given a drug under study)?
  • What is the life expectancy of the population in Gurgaon?

  However this technique/model(s) are now used for variety of tasks such as to determine machine failure time, customer churn duration etc. These models are different from the regression models in the following way:

  • Survival analysis is specially designed to handle data censorship
  • In survival analysis we aim to derive the survival/hazard functions unlike lifespan values for individual samples as in a regression setup.

• Brief Description and Contents to be covered

  The aim of this lighting talk is:

  • Quick understanding of the motivation behind using survival analysis and it's USP.
  • Touch upon the basic concepts of data censorship
  • Couple of coding examples to see how can we use survival analysis approach to handle censored data

• Pre-requisites for the talk

  Basic concepts of probability (maximum likelihood estimation in particular) and working knowledge of Python and Jupyter are pre-requisite to attend this talk

• Time required for the talk

  15 minutes

• Link to slides

  Survival Analytics - These slides had been designed for 1 hour long session. I need to make necessary changes to make it suitable for 15 minute session. Suggestions on this would be helpful.

• Will you be doing hands-on demo as well?

  Yes

• Link to ipython notebook (if any)

  Github Repo containing Notebooks

• About yourself

  I am a Data Science Analyst at Deloitte Consulting LLP. I analyse data from clients across industries to provide actionable insights, build predictive/analytical models to help them tap their data reservoirs for improved business decisions. I use Python extensively for my day to day machine learning and data analysis tasks. I am a postgraduate from CSRE, IIT Bombay in Geoinformatics and Natural Resources Engineering. Majority of my work at CSRE was focused in satellite image processing. I love open source and I love the open source community even more.

• Are you comfortable if the talk is recorded and uploaded to PyData Delhi's YouTube channel ?

  Yes I would love to have that :)

• Any query ?

  Suggestions on how to make my content crisp for 15 minutes would be most welcome.

[shagunsodhani]

Looks good to me. The slides are quite dense. I understand that you are going to trim them for 15 min but I am referring to the text density :) Since it is a lightening talk, I would suggest keeping the slides to the minimal and focusing more on practical aspects.

[thisisashukla]

@shagunsodhani thanks for the suggestion. I'll make the changes.

[thisisashukla]

@shagunsodhani @MSanKeys963 i have updated the deck and notebook (new URLs above). Made them crisp for 15 minutes. Your feedback would be helpful. Thanks

[shagunsodhani]

Thanks @thisisashukla The slides look clean. My only feedback is to go a little slow on the slides with the equations as sometimes people are intimidated by maths :)

[thisisashukla]

@shagunsodhani yeah I'll take care with the slides with equation.

[thisisashukla]

@MSanKeys963 is this talk confirmed?

[MSanKeys963]

Yeah, it's confirmed.

[thisisashukla]

thanks a ton @MSanKeys963 and @shagunsodhani . i'll try my best to do justice with the time given and the topic

[greatsharma]

@thisisashukla can you please share the entire original slides for indepth reading

[thisisashukla]

sure @greatsharma here is the link https://drive.google.com/file/d/1Om8kkO5ZR8W7_qZriMhy3pCJ8ZvqZLAC/view?usp=sharing