Inspiration
Predicting Customer Purchases for an E-Commerce Company About the Project This project focuses on building a predictive model to determine whether a customer will make their next purchase based on their profile and purchasing behavior. The goal is to help an E-Commerce company optimize marketing efforts and customer retention strategies.
Inspiration With the rise of online shopping, businesses want to predict customer behavior to improve engagement. By leveraging machine learning, we can help businesses target the right customers and increase sales efficiently.
How We Built It Data Exploration & Cleaning
Loaded the dataset and checked for missing values. Handled missing data with median imputation for numerical features. Encoded categorical features using Label Encoding. Feature Engineering & Preprocessing
Scaled numerical data. Ensured the test dataset followed the same transformations as training data. Model Selection & Training
Used a Random Forest Classifier for training. Split data into training and validation sets. Tuned hyperparameters and evaluated model performance. Making Predictions & Submission
Processed the test dataset. Generated predictions (0 = No purchase, 1 = Purchase). Created and exported the submission.csv file. Challenges Faced Handling unseen categories in the test dataset that were not present in training. Balancing the dataset, as positive cases (customers making a purchase) were less frequent. Improving recall to capture more potential customers while keeping false positives low. Key Learnings The importance of feature preprocessing for real-world datasets. How to handle categorical encoding safely to avoid errors. Evaluating classification models using precision, recall, and F1-score instead of just accuracy.
Log in or sign up for Devpost to join the conversation.