# How to predict customer churn (attrition) one month after start date? [closed]

I am trying to understand the strength of impact of variables on customer churn (attrition). I want to predict the probability that a customer will churn after time period t (after 1 months, after 3 months) from start date of service.

My intention is not to predict how many customers will churn in next month. Instead, I want predict outcome given a customer start date and other variable. I need to know the probability of churn after one month from start date (customer will stay for one month before churn). Can I use linear regression or survival analysis methods? Explaining the predictive ability of variables is also important to me.

My data looks like this, i have several independent variables. date format is mm/dd/yyyy

id  start_date var1 var2   end_date
1   1/1/2018    30   2     2/15/2018
2   1/24/2018   5    9     2/10/2018
3   3/2/2018    10   3     4/26/2018


customer 1 and 3 stayed with us for more than 1 months. whereas customer 2 churned before spending one month with us. start date and end date can be anything in last 2 years. my target is customer will spend at least one month in our service.

## closed as unclear what you're asking by Michael Chernick, Carl, kjetil b halvorsen, mdewey, Jeremy MilesSep 12 '18 at 5:04

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

• Without presenting data, it is hard to offer an opinion. Would you please present some data? – Carl Sep 9 '18 at 0:23
• I have updated the questions with some sample data. – Gowtham M Sep 9 '18 at 1:29
• Consider making a churned column with 0 (no churn) and 1 (churned). If the customer churns, does the date then represent the date of churning? What are var1 and var2? If you want to make a physical argument then we need to know what these variables represent. Put in enough data for us to make sense of it, (lots more). – Carl Sep 9 '18 at 2:21
• encode 0 and 1 and build a classification model like logistic regression? – Gowtham M Sep 9 '18 at 2:24
• Don't know yet. Put in more info. – Carl Sep 9 '18 at 2:27

## 1 Answer

If you have the end-date for all customers, then as far as I can see, you don't have censoring. Hence, you could treat the time a continuous variable (possibly consider a transformation, such as the log if it has a non-symmetric distribution), and use a linear regression model.