In today's world, it is becoming increasingly important for companies to reach the right people from their customer pool to market their products and services. This is also true for banking and financial institutions, as it can be a costly and time-consuming process to reach the right people from a large customer base. In line with this, institutions have increased their machine learning-based models using demographic and financial data of customers who have purchased products and services. The interest in this technology has increased as these models have both successfully identified the right people and reduced the time and cost of money. Machine learning-based models are frequently used in this direction.

In this study, machine learning prediction models have been established on whether customers who prefer the financial institution in Portugal will take a term deposit. Thus, the institution will identify existing customers who are more likely to subscribe to a term deposit and focus on these customers with marketing efforts. In the study, the data set containing the information of whether the financial institution in Portugal wants a term deposit account by calling its customers was used. The modeling processes were performed automatically with data preprocessing steps and the “DataBlender” automatic machine learning tool was used for this. As a result of the study, the “LightGBM” model was the most successful model among the models. As a result, it was observed that the success criteria of tree-based models are close to each other. The effective variables on customers' term deposit taking behavior were found as a result of the model, and the customer's features were ranked by impact.