mpc_score

empulse.metrics.mpc_score

Maximum Profit Measure for Customer Churn (MPC).

MPC presumes a situation where identified churners are contacted and offered an incentive to remain customers. Only a fraction of churners accepts the incentive offer, described by a constant accept_rate. For detailed information, consult the paper [1].

See empc_score for a stochastic version of this metric.

The MPC is defined as [1]:

\[CLV (\gamma (1 - \delta) - \phi) \pi_0 F_0(T) - CLV (\delta + \phi) \pi_1 F_1(T)\]

The MPC requires that the churn class is encoded as 0, and it is NOT interchangeable. However, this implementation assumes the standard notation (‘churn’: 1, ‘no churn’: 0).

Methods

__call__(y_true, y_score, *, accept_rate=0.3, clv=200, incentive_cost=10, contact_cost=1)

Compute the maximum profit that can be achieved by a classifier at its optimal decision threshold.

optimal_threshold(y_true, y_score, *, accept_rate=0.3, clv=200, incentive_cost=10, contact_cost=1)

Compute the classification threshold that maximizes the profit.

optimal_rate(y_true, y_score, *, accept_rate=0.3, clv=200, incentive_cost=10, contact_cost=1)

Compute the predicted positive rate (fraction of the customer base that should be targeted) at which the maximum profit is achieved.

References

[1] (1,2)

Verbraken, T., Verbeke, W. and Baesens, B. (2013). A Novel Profit Maximizing Metric for Measuring Classification Performance of Customer Churn Prediction Models. IEEE Transactions on Knowledge and Data Engineering, 25(5), 961-973.

Examples

from empulse.metrics import mpc_score

y_true = [0, 1, 0, 1, 0, 1, 0, 1]
y_score = [0.1, 0.2, 0.3, 0.4, 0.5, 0.7, 0.8, 0.9]
mpc_score(y_true, y_score, accept_rate=0.3, clv=200, incentive_cost=10, contact_cost=1)