Closed maria-ferreira-olx closed 2 months ago
That's actually probably because the fitted model has parameters that force P_alive
to be very close to either 0
or 1
.
Try using the conditional_probability_alive()
model's method to inspect various customers' P_alive
. There is also a calculate_alive_path()
function in the utils.py
file that might be useful — it uses conditional_probability_alive()
and is the function being called inside the plot_alive_path()
.
Also, please post code snippets and graphs, not screenshots, it's much easier for others to debug.
The code works ok, but when plotting P_alive it doesn't show the p, only the purhcases it works on other dataset, but not following this one https://github.com/susanli2016/Machine-Learning-with-Python/blob/master/CLV_Online_Retail.ipynb