Let’s learn about Survival Analysis. By this name , we can get idea about the meaning of survival. In our predictive analytical problems , we are dealing with various situations which are important events along a customer’s journey and based upon the concept of time to event. Some events like a car crash , a stock market crash etc.
There are many statistical modeling techniques which can be chosen for analysis like regression , decision trees, random forests, naive bayes and neural networks. Here involve the concepts of time-independent data and censoring.
What is time-dependent data? Any event that measures a variable change over time like age can be treated as time-dependent variables and are handled by survival analysis method.
Let’s know about censoring. Censoring is used to describe data which is partially known. For an observation, data is captured between starting and ending periods. But some of the information is captured before a study has begun or after a study has ended. This type of information is called as censored information. Censored data can be left or right censored.
For a marketing context, studies begin with customers already in place, and without any knowledge of how they were acquired. But not all customers start at the same time. When a customer has started prior to the beginning of a study , some customer attributes can be referred to as left censored. A customer is considered right censored if a study ends before the customer terminates. It also includes some other special situations, such as when a customer is still active at the end of the study, or is lost due to events not related to the study’s variables.