A series of direct marketing campaigns using phone calls were conducted by the Portuguese banking institutions to promote a term deposit to its customers. Here, a term deposit is a financial product where a customer deposits an amount with the bank for a fixed period at a predetermined interest rate. One of the main objectives of these marketing campaigns is to encourage clients to subscribe to a term deposit and this can be a source of revenue for the bank. The UCI Bank Marketing Dataset is a valuable resource for financial institutions to make data driven decisions in their marketing efforts. The objective is to build source-code level predictive models (classification) to help the bank optimize its marketing campaigns and thereby improve subscription rates. This further extends to predicting whether a client will subscribe to a term deposit and gain insights into the efficacy of various marketing strategies.
- Explored bank marketing campaign data to understand factors included such as age, job, loan default, etc. and their relation to the output of the campaigns.
- Preprocessed data to handle missing/NaN values, scale data using standardization and one-hot encoding, and improve data imbalance by implementing the synthentic minority oversampling technique (SMOTE).
- Leveraged random forest classifier to select top 10 optimal features for model training.
- Scripted logistic regression, SVM, and gaussian naive bayes to predict customer's output from such campaigns on new data.
- Python
- Jupyter Notebook