AI Forecasting App

Directory Structure Update (July 2025)

The helpers folder and global_helpers.py have been moved into a new server_scripts folder for better modularity. Please update your imports accordingly:

• Use from server_scripts.global_helpers import ... instead of from global_helpers import ...
• Use from server_scripts.helpers import ... for helper modules.

A PyShiny application for time series forecasting using various AI and statistical models.

Features

• Upload CSV data or use sample data from MongoDB
• Interactive data visualization and editing
• Multiple forecasting models:
  • Prophet
  • Auto ARIMA
  • LSTM (Deep Learning)
  • AutoML (H2O)
  • ARFIMA
• Forecast metrics and evaluation
• Modern, responsive UI

Installation

1. Clone this repository
2. Install the required dependencies:

pip install -r requirements.txt

3. Make sure you have MongoDB installed and running (optional, for sample data functionality)

Usage

Run the application:

shiny run app.py

Then open your browser and navigate to http://localhost:8000

Data Format

The application expects CSV files with at least one time column and one or more numeric columns for forecasting. You can download a template from the application.

Models

Prophet

Facebook's Prophet model for time series forecasting with support for yearly, weekly, and daily seasonality.

Auto ARIMA

Automatic ARIMA model selection with support for seasonal components.

LSTM

Long Short-Term Memory neural network for sequence prediction.

AutoML

H2O AutoML for automated machine learning model selection and training.

ARFIMA

AutoRegressive Fractionally Integrated Moving Average model.

Author

Soumyadipta Das

Version

0.03.3