vinayakdnrdd/superkart-sales-xgb-model

SuperKart Sales Forecast Model (XGBoost Regressor)

This repository contains a serialized scikit-learn Pipeline (preprocessing + XGBoost regressor) trained on the SuperKart dataset.

Target

• Product_Store_Sales_Total

Training data

• Dataset repo: vinayakdnrdd/superkart-sales-data
• processed/train.csv
• processed/test.csv

Evaluation (test split)

• RMSE: 231.3869
• MAE: 73.3437
• R²: 0.9531

Best Hyperparameters

{
  "model__subsample": 0.85,
  "model__reg_lambda": 1.0,
  "model__n_estimators": 800,
  "model__max_depth": 6,
  "model__learning_rate": 0.03,
  "model__colsample_bytree": 0.85
}

How to use

import joblib
from huggingface_hub import hf_hub_download

model_path = hf_hub_download('vinayakdnrdd/superkart-sales-xgb-model', filename='superkart_sales_model.joblib')
model = joblib.load(model_path)
y_pred = model.predict(X)  # X: pandas DataFrame with training feature columns