13.6 AN APPLICATION OF POISSON REGRESSION MODELING
We will use the churn data set to build a model that estimates the number of customer service calls based on whether a customer churned. Our response variable is an integer‐valued variable, which is why we use Poisson regression instead of linear regression for this estimation.
The structure of our Poisson regression model will be
![equation](https://static.packt-cdn.com/products/9781119526810/graphics/images/c13-disp-0015.png)
The result of the regression analysis is given in Figure 13.5. Using the coefficients given above, we can build the Poisson regression model
![equation](https://static.packt-cdn.com/products/9781119526810/graphics/images/c13-disp-0016.png)
![No alt text required.](https://static.packt-cdn.com/products/9781119526810/graphics/images/c13f005.gif)
Figure 13.5 Python Poisson regression results for predicting number of customer service calls.
Now, how do we interpret the Poisson regression coefficients? When used as the exponent of e, the regression coefficient describes the estimated multiplicative change in the response variable when the coefficient's predictor variable increases by one. In our case...