usman-tech-ali/stockforecast-traditional-models

StockForecast Traditional Models

This repository contains traditional time series forecasting models for financial data, part of the StockForecast AI project for CS4063 NLP Assignment 2.

Models Included

Moving Average Forecaster

• Algorithm: Simple Moving Average with configurable window
• Window Size: 5 (default)
• Use Case: Trend following and baseline performance
• Performance: RMSE=2.45, MAE=1.89, MAPE=1.85%

ARIMA Forecaster

• Algorithm: AutoRegressive Integrated Moving Average
• Order: (1,1,1)
• Use Case: Time series with trend and seasonality
• Performance: RMSE=2.12, MAE=1.67, MAPE=1.64%

Usage

import joblib
from huggingface_hub import hf_hub_download

# Download models
ma_model_path = hf_hub_download(repo_id="usman-tech-ali/stockforecast-traditional-models", filename="moving_average_model.pkl")
arima_model_path = hf_hub_download(repo_id="usman-tech-ali/stockforecast-traditional-models", filename="arima_model.pkl")

# Load models
ma_model = joblib.load(ma_model_path)
arima_model = joblib.load(arima_model_path)

# Make predictions
ma_prediction = ma_model.predict(steps=5)
arima_prediction = arima_model.predict(steps=5)

Dataset

Trained on financial OHLCV data with technical indicators.

Citation

@software{stockforecast_ai_2025,
  title={StockForecast AI: Complete Financial Forecasting Application},
  author={Usman Ali},
  year={2025},
  url={https://github.com/usman-tech-ali/stock-forecast-app}
}