Time series forecasting on Superstore sales data using ARIMA and Holt–Winters models. Includes data preprocessing, decomposition, model comparison, and 12-month sales forecast. Holt–Winters (Multiplicative Trend) selected for best accuracy.

Unsupervised customer segmentation on the UCI Online Retail II dataset using RFM features and KMeans. Includes full pipeline: data cleaning, feature engineering, outlier handling, model selection (Elbow & Silhouette), and actionable segment insights for targeted marketing.