Welcome to ts-panel-forecasting-baselines! This application helps you forecast daily sales across different stores and items. It includes clean validation methods, strong statistical baselines, and automated models, making it easy to generate accurate predictions.
- Multiple Statistical Models: Use SARIMAX, TBATS, and ARIMA for your forecasting needs.
- Automated Forecasting: Implement AutoTS for an easy setup.
- User-Friendly Notebooks: Work with straightforward Jupyter notebooks for all analyses.
- Validation Methods: Get reliable validation using holdout and rolling-origin backtesting techniques.
- Optional Enhancements: Use Prophet, Darts, or NeuralProphet for additional forecasting options.
- Key Metric: Measure performance with SMAPE for accurate forecasting evaluations.
To use ts-panel-forecasting-baselines, you need:
- Operating System: Windows, macOS, or any Linux distribution.
- Python Version: Install Python 3.7 or higher.
- Memory: At least 4 GB of RAM.
- Storage: Minimum of 500 MB available disk space for installation.
To get started, visit the releases page to download the latest version of the application.
- Access the Releases Page: Click the link above to navigate to the releases section of the repository.
- Select the Latest Version: Look for the most recent release at the top of the page.
- Download the File: Choose the appropriate file for your operating system and click on it to download.
- Install the Application: Open the downloaded file and follow the installation instructions.
Once you have installed the application, follow these steps to start forecasting:
-
Launch Jupyter Notebooks:
- Open the Jupyter Notebook interface on your machine. You can do this via the command prompt by typing
jupyter notebook.
- Open the Jupyter Notebook interface on your machine. You can do this via the command prompt by typing
-
Open a Notebook:
- Inside the Jupyter interface, navigate to the directory where you installed the application. Open one of the available notebooks to begin.
-
Load Your Data:
- Prepare your dataset with daily sales information for different stores and items. Import this data into the notebook as guided in the notebook's comments.
-
Choose a Model:
- Select the forecasting model you want to use. You can choose from several options like SARIMAX, TBATS, or ARIMA, as demonstrated in the provided notebooks.
-
Run the Analysis:
- Execute the code blocks in the notebook step-by-step. Each section will guide you through loading data, training the model, and generating predictions.
-
Evaluate Results:
- Review the outputs and predictions from your model. Compare results using the SMAPE metric to assess the accuracy of your forecasts.
- Data Quality: Ensure your dataset is clean and free from errors for the best forecasting accuracy.
- Model Experimentation: Try different models to see which provides the best results for your data.
- Review Documentation: Check the comments within the notebooks for detailed annotations and further instructions.
If you encounter any issues or have questions, feel free to open an issue on the GitHub repository. The community and the maintainers are here to help you.
This project includes various topics related to time-series analysis such as:
- ARIMA
- AutoTS
- Backtesting
- Darts
- Forecasting
- NeuralProphet
- Notebooks
- Optuna
- Panel Data
- Prophet
- Python
- Rolling-Origin
- SARIMAX
- SMAPE
- TBATS
- Time-Series
Explore these topics to enhance your understanding and capabilities in forecasting sales data.
For deeper insights into the methodologies and algorithms used in this project, consider exploring the following:
Happy forecasting!