This project demonstrates Time Series Forecasting using statistical and machine learning techniques. The notebook explores different forecasting models and evaluates their performance on time-based datasets.

The goal of this project is to:

• Perform time series data preprocessing
• Generate time-based features
• Apply statistical forecasting models
• Evaluate model performance
• Visualize forecasting results

• Python
• Pandas
• NumPy
• Matplotlib
• StatsForecast
• UtilsForecast
• Jupyter Notebook

git clone cd

python -m venv venv Activate environment:

Windows: venv\Scripts\activate

Linux / Mac: source venv/bin/activate

pip install -r requirements.txt

Run the Jupyter notebook: jupyter notebook

Open: time_series_forecasting.ipynb

AutoARIMA

Naive Forecasting

Seasonal Naive

Historic Average

Window Average

The project evaluates forecasting accuracy using metrics from UtilsForecast.

Add Deep Learning models (LSTM / Transformer)

Hyperparameter tuning

Deployment as API or Web App

Real-time forecasting pipeline

Anurag Prajapati

Give it a star on GitHub ⭐