1

I am trying to predict - Number of days it takes for a customer to make the second purchase. Sometimes the customer comes back in 2,5,6,10.... days and sometimes the customer does not come back which is indicated by NULL in my data. Since my Target Variable can be NULL and some other values, should I think about using Linear Regression? Do you think Using Logistic Regression is better approach?

Also, if my target variable is right skewed should I apply transformations? Will it impact the way I analyze my results

whuber
  • 281,159
  • 54
  • 637
  • 1,101
Pratibha UR
  • 11
  • 1
  • The customers that 'don't come back' might come back after you've stopped watching. So it may be helpful to look at this as a matter of [censoring](https://stats.stackexchange.com/questions/133347/ml-estimate-of-exponential-distribution-with-censored-data). – BruceET Oct 03 '19 at 02:42

1 Answers1

0

The tool you are looking for is survival analysis. Its main use is in clinical analyses, whether people die sooner or later after some intervention. Of course, sometimes people don't die at all during the course of the experiment. In your case, the analogue of a patient's death would be a customer's return.

Stephan Kolassa
  • 95,027
  • 13
  • 197
  • 357