What is supermarket-sales-prediction-using-machine-learning?

This is a data science capstone project that builds a machine learning model to predict supermarket transaction sales from historical data. It was completed as part of a data science and machine learning internship program, and it walks through the full process a beginner data scientist would follow, from cleaning the data to producing a working prediction model. The project uses a public Kaggle dataset containing 1,000 supermarket transactions with details like branch, city, customer type, gender, product line, price, quantity, payment method, date, time, and a satisfaction rating. It analyzes this data to understand what factors influence sales, including patterns by product line, branch, day of the week, and time of day. Before training any models, the notebook prepares the data by converting the date into separate year, month, day, and hour fields, converting category names into a machine-readable format, and removing a few columns that would let the model cheat by seeing information it should not have access to ahead of time. Four different regression models were trained and compared: Linear Regression, Decision Tree, Random Forest, and Gradient Boosting. The Random Forest model performed best, explaining about 99.8 percent of the variation in sales values based on the accuracy metrics reported. The README also lists a few business takeaways drawn from the data, such as weekend sales tending to run higher than weekday sales and e-wallet being the most common payment method. The project is built in Python using Pandas, NumPy, Matplotlib, Seaborn, and Scikit-learn, and can be run either as a local Jupyter Notebook or uploaded to Google Colab. This would suit someone learning data science who wants a complete worked example of a regression project, from raw data to a trained model and written business insights.