Srikanth-Karthi/ecommerce-spending-predictor

E-commerce Spending Predictor

A Linear Regression model that predicts yearly customer spending based on user behavior metrics from an e-commerce platform.

Model Description

• Model Type: Linear Regression (scikit-learn)
• Task: Tabular Regression
• Language: Python
• License: Apache 2.0

Features

Feature	Description
Avg. Session Length	Average session length in minutes
Time on App	Time spent on mobile app in minutes
Time on Website	Time spent on website in minutes
Length of Membership	Years of membership

Target Variable

• Yearly Amount Spent - Total yearly spending in USD

Performance Metrics

Metric	Value
Mean Absolute Error (MAE)	8.43
Mean Squared Error (MSE)	103.92
Root Mean Squared Error (RMSE)	10.19

Quick Start

from huggingface_hub import hf_hub_download
import joblib

# Download and load model
model_path = hf_hub_download(
    repo_id="Srikanth-Karthi/ecommerce-spending-predictor",
    filename="model.joblib"
)
model = joblib.load(model_path)

# Predict
features = [[33.0, 12.0, 37.0, 3.5]]
# [Avg Session Length, Time on App, Time on Website, Membership Years]

prediction = model.predict(features)
print(f"Predicted Yearly Spending: ${prediction[0]:.2f}")

Training Details

• Algorithm: Linear Regression
• Library: scikit-learn
• Train/Test Split: 70/30
• Random State: 42

Use Cases

• Customer lifetime value prediction
• Marketing budget allocation
• Customer segmentation
• Revenue forecasting

Limitations

• Model assumes linear relationships between features and target
• Trained on a specific e-commerce dataset (500 samples)
• May not generalize to different industries or customer bases

Citation

@misc{ecommerce-spending-predictor,
  author = {Srikanth-Karthi},
  title = {E-commerce Spending Predictor},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/Srikanth-Karthi/ecommerce-spending-predictor}
}