Learn Predicting Customer Churn | Growth, Marketing, and Customer Insights

Predicting which customers are likely to leave your service—known as customer churn—is essential for startup growth. Churn directly impacts your revenue and can signal deeper issues with your product or customer experience. By identifying customers at risk of leaving, you can take targeted action to improve retention and reduce costly turnover.


              123456789101112131415161718
            
import pandas as pd
from sklearn.linear_model import LogisticRegression

# Hardcoded customer data
data = {
    "monthly_usage": [120, 80, 200, 50, 300, 60, 90, 250],
    "num_complaints": [1, 3, 0, 5, 0, 4, 2, 1],
    "churned": [0, 1, 0, 1, 0, 1, 1, 0]
}
df = pd.DataFrame(data)

# Features and labels
X = df[["monthly_usage", "num_complaints"]]
y = df["churned"]

# Fit logistic regression
model = LogisticRegression()
model.fit(X, y)

A logistic regression model is a popular choice for predicting binary outcomes like churn (yes/no). It estimates the probability that each customer will churn based on their features, such as monthly_usage and num_complaints. After training, you can interpret the model's predictions as the likelihood of a customer leaving, which helps you prioritize retention efforts.


              12345678910111213
            
# Predict churn for new customers
new_customers = pd.DataFrame({
    "monthly_usage": [70, 220],
    "num_complaints": [4, 0]
})
predictions = model.predict(new_customers)
probabilities = model.predict_proba(new_customers)[:, 1]

# Summarize results
for i, (pred, prob) in enumerate(zip(predictions, probabilities)):
    print(
        f"Customer {i+1}: Predicted churn = {bool(pred)}, Probability = {prob:.2f}"
    )

Everything was clear?

Thanks for your feedback!

Section 3. Chapter 4

Ask AI

Ask anything or try one of the suggested questions to begin our chat

Suggested prompts:

Can you explain how the model decides if a customer will churn?

What do the probability values mean in the predictions?

How can I improve the accuracy of this churn prediction model?

Swipe to show menu


              123456789101112131415161718
            
import pandas as pd
from sklearn.linear_model import LogisticRegression

# Hardcoded customer data
data = {
    "monthly_usage": [120, 80, 200, 50, 300, 60, 90, 250],
    "num_complaints": [1, 3, 0, 5, 0, 4, 2, 1],
    "churned": [0, 1, 0, 1, 0, 1, 1, 0]
}
df = pd.DataFrame(data)

# Features and labels
X = df[["monthly_usage", "num_complaints"]]
y = df["churned"]

# Fit logistic regression
model = LogisticRegression()
model.fit(X, y)


              12345678910111213
            
# Predict churn for new customers
new_customers = pd.DataFrame({
    "monthly_usage": [70, 220],
    "num_complaints": [4, 0]
})
predictions = model.predict(new_customers)
probabilities = model.predict_proba(new_customers)[:, 1]

# Summarize results
for i, (pred, prob) in enumerate(zip(predictions, probabilities)):
    print(
        f"Customer {i+1}: Predicted churn = {bool(pred)}, Probability = {prob:.2f}"
    )

Everything was clear?

Thanks for your feedback!

Section 3. Chapter 4

Predicting Customer Churn

1. What is customer churn?

2. How can predictive modeling help reduce churn?

3. Which scikit-learn model is commonly used for binary classification tasks like churn prediction?

Predicting Customer Churn

1. What is customer churn?

2. How can predictive modeling help reduce churn?

3. Which scikit-learn model is commonly used for binary classification tasks like churn prediction?