From data to action: Empowering COVID-19 monitoring and forecasting with intelligent algorithms

From data to action: Empowering COVID-19 monitoring and forecasting with intelligent algorithms

COVID-19 Time Series Forecasting

This repository implements COVID-19 time series forecasting using advanced techniques inspired by research from the paper From data to action: Empowering COVID-19 monitoring and forecasting through advanced analytics. It includes methods such as ARIMA, BiLSTM, GMDH, Genetic Algorithm, and TSK Fuzzy Logic for robust and accurate forecasting of pandemic trends.

Please cite:

Charles, Vincent, et al. "From data to action: Empowering COVID-19 monitoring and forecasting with intelligent algorithms." Journal of the Operational Research Society 75.7 (2024): 1261-1278.

Features

• Forecasting Models:

  • ARIMA: Auto-Regressive Integrated Moving Average for statistical forecasting.
  • BiLSTM: Bidirectional Long Short-Term Memory networks for deep learning-based predictions.
  • GMDH: Group Method of Data Handling for self-organizing polynomial regression.
  • Genetic Algorithm: Evolutionary optimization for parameterized forecasting.
  • TSK Fuzzy Logic: Fuzzy inference with Takagi-Sugeno-Kang rules for uncertainty modeling.

• Preprocessing:

  • Sliding window feature extraction for time series.
  • Min-Max normalization for feature scaling.

• Visualization:

  • Historical vs. forecasted data plots.
  • Comparative performance metrics across models.

---

📊 Methodology

1Forecasting Techniques

• ARIMA:

  • A classical statistical model for forecasting time series data.
  • Best suited for univariate data with stationary patterns.

• BiLSTM:

  • Deep learning-based model that captures temporal dependencies in both forward and backward directions.
  • Requires large training datasets for optimal performance.

• GMDH:

  • A polynomial regression model with self-organizing capabilities.
  • Automatically selects the best subset of features and model complexity.

• Genetic Algorithm:

  • Optimizes regression weights by evolving solutions over generations.
  • Suitable for parameter tuning in complex forecasting models.

• TSK Fuzzy Logic:

  • Uses Gaussian membership functions for fuzzification.
  • Implements fuzzy IF-THEN rules with linear consequents for robust predictions.

---

• Download the dataset by "john hopkins university":
• https://github.com/CSSEGISandData?tab=repositories
• https://github.com/CSSEGISandData/COVID-19_Unified-Dataset
• Link to papers:
• https://www.tandfonline.com/doi/full/10.1080/01605682.2023.2240354#abstract
• http://dx.doi.org/10.6084/m9.figshare.14396258