ML model + retention overview
Player data
Loading…
❗The whole data set has available date from 2023-01-01 to 2024-03-31.
Model performance
The overall proportion of correct predictions (both retained and churned players) out of all predictions made.
The proportion of correctly identified churned players out of all players predicted to churn. A lower precision suggests some false positives in churn prediction.
This very high recall indicates the model is very good at identifying actual churned players, capturing about 92% of all truly churned players.
This is the harmonic mean of precision and recall, providing a balanced measure of the model's performance.
The combination of metrics suggests a model that is very sensitive to detecting churn but may be overestimating churn risk in some cases. This could be preferable if the cost of missing a potentially churning player is higher than the cost of wrongly flagging a player as at risk of churning.
Personalised retention system
Possible future usage for slack or google integration.