In my project, I aimed to predict whether clients would subscribe to insurance based on demographic and marketing data. I cleaned and analyzed the data, transforming it into a suitable format for machine learning. After training several models, I found that the Gradient Boosting model performed best, achieving a high ROC AUC score of 0.986 and an accuracy of 77.3%. I discovered that the duration of the call was the most significant predictor of subscription. This model can effectively help identify potential insurance subscribers.