top of page

Loan Prediction

Model to Predict Client Subscription to Term Deposit

Problem Statement:

  • A bank needs to identify customers who are likely to subscribe to Term Deposit. This will help the bank improve its strategies and optimize marketing campaigns by targeting the right customers.

Objectives:

  • Identify factors affecting propensity to subscribe to Term Deposit

  • Build a predictive model to identify and segment customers based on their likelihood to get the product

  • Deploy the application on a cloud platform so that it is accessible to users

Summary:

  • The project evaluates and optimizes a marketing campaign dataset to predict customer subscription to a product which is Term Deposit. The dataset includes various features such as age, job type, marital status, education level, and financial information with the target variable being whether a customer subscribed ('yes') or not ('no'). The project uses feature engineering to enhance predictive capabilities of machine learning models. Two models are employed: Decision Tree classifier and Artificial Neural Network (ANN). Both the Decision Tree and ANN models are evaluated for accuracy, precision, recall, and F1 score.

  • The project concludes with a comparative analysis of the Decision Tree and ANN models, emphasizing their strengths and weaknesses in predicting customer subscriptions. The performance of the ANN and decision tree models is nearly identical. But since the decision tree is easier to understand and more straightforward, this will be the final model. This comprehensive approach ensures a thorough understanding of the marketing campaign's effectiveness and facilitates data-driven decision-making for future campaigns.

Methodology:

  • Decision Trees: These are powerful models that are easy to understand and interpret. In this model, the decision tree helps identify significant features and their relationships with the target variable (subscription to term deposit). Decision trees are particularly useful for feature importance analysis, and they assist in creating a clear decision-making path based on the dataset's characteristics. The visualization of the decision tree provides transparency in understanding the rules that lead to a positive or negative outcome. This simplicity is advantageous when the interpretability of the model is crucial.

  • Artificial Neural Networks (ANN): These are highly flexible and can capture complex relationships within the data. In this model, the ANN's ability to learn intricate patterns and non-linear dependencies between features makes it suitable for scenarios where decision boundaries are complex. ANNs are capable of automatically extracting relevant features and understanding intricate patterns that might be challenging for a decision tree to capture. They offer a higher level of predictive accuracy by modeling intricate relationships between input features and the target variable.

  • Combining Decision Tree and ANN: In this project, both models were used in tandem to leverage the interpretability of the decision tree for feature analysis and rule extraction, while benefiting from the ANN's capacity to capture intricate patterns in the data. This hybrid approach allowed for a well-rounded understanding of the underlying factors influencing the subscription to term deposit, providing both transparency and predictive accuracy.

bottom of page