{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/186ef8fa067c4650ae3a71975ac03d20\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/186ef8fa067c4650ae3a71975ac03d20-36b2b3895b692754.gif","duration":714.832,"title":"8-Week State Sales Forecasting API Build 🤖","description":"In this Loom, I built a production style forecasting pipeline to predict the next 8 weeks of sales for each state. I preprocess and resample data to a weekly interval, handle missing values and duplicates, then do feature engineering with lag features, rolling mean and std, temporal features like month and quarter, and a holiday flag. I train four models per state, SARIMA, ARIMA, Profit XGBoost, and LSTM, evaluate with RMSE, MAE, and MAPE, and automatically select the best validation performer. I save trained models with joblib and serve forecasts through a FastAPI REST API with Swagger. No action is requested from viewers."}