A machine learning model was developed for a telecom company, effectively predicting customer churn with an accuracy of 0.88. The model utilized customer data including demographics, services, account details, and helped formulate a data-centric strategy to enhance the existing customer retention program.
The model's performance evaluation revealed essential factors such as monthly charges, tenure, and contract type playing a crucial role in predicting churn. Interesting findings suggested that customers with a month-to-month contract were proportionally more likely to churn, impacting the churn prediction model's sensitivity and specificity indicators.
Insights drawn from the model formed the basis for strategic decision-making, particularly in resource allocation for retention efforts. Particularly, it highlighted the cost-effectiveness of not targeting long-term customers based on their low churn probability, thereby maximizing the impact of the retention program.
.svg)
.png)
.png)
.png)
.png)
.png)
.png)