Logistic regression algorithm

Let's dig into the internals and implement a logistic regression algorithm. Since statsmodels's logit() function is very complex, you'll stick to implementing simple logistic regression for a single dataset.

Rather than using sum of squares as the metric, we want to use likelihood. However, log-likelihood is more computationally stable, so we'll use that instead. Actually, there is one more change: since we want to maximize log-likelihood, but minimize() defaults to finding minimum values, it is easier to calculate the negative log-likelihood.

The log-likelihood value for each observation is $$ log(y_{pred}) * y_{actual} + log(1 - y_{pred}) * (1 - y_{actual}) $$

The metric to calculate is the negative sum of these log-likelihood contributions.

The explanatory values (the time_since_last_purchase column of churn) are available as x_actual. The response values (the has_churned column of churn) are available as y_actual. logistic is imported from scipy.stats, and logit() and minimize() are also loaded.

This exercise is part of the course

Intermediate Regression with statsmodels in Python

View Course

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Complete the function
def calc_neg_log_likelihood(coeffs):
    # Unpack coeffs
    ____, ____ = ____
    # Calculate predicted y-values
    y_pred = ____
    # Calculate log-likelihood
    log_likelihood = ____
    # Calculate negative sum of log_likelihood
    neg_sum_ll = ____
    # Return negative sum of log_likelihood
    return ____

# Test the function with intercept 10 and slope 1
print(calc_neg_log_likelihood([10, 1]))