Time Series Prediction Service

A RESTful API service for time series forecasting built with FastAPI and Prophet. This service provides scalable, customizable time series predictions through a simple HTTP interface.

Features

• RESTful API: Easy-to-use HTTP endpoints for time series forecasting
• Customizable Forecasting: Extensive model parameter customization options
• Rich Output: Returns predictions with confidence intervals and seasonal components

Components

Backend Service (app.py)

• FastAPI application providing the REST API
• Prophet model integration for forecasting
• Customizable model parameters

Example Request (example_req.py)

• CSV data loading and preprocessing
• API interaction utilities
• Results saving and formatting
• Basic forecast summary statistics

Installation

1. Clone the repository
2. Create a virtual environment (for example with venv):

python -m venv venv

3. Install requirements:

pip install -r requirements.txt

Usage

Starting the Server

python app.py

The server will start at http://localhost:8000

Making Predictions

1. Prepare your time series data in CSV format with date and value columns
2. Use the client script:

python example_req.py

API Endpoints

• POST /forecast/: Create time series forecasts
• GET /parameters/default: Get default model parameters

Example Request

{
    "data": {
        "dates": ["2023-01-01", ...],
        "values": [100, ...]
    },
    "periods": 30,
    "model_parameters": {
        "changepoint_prior_scale": 0.08,
        "seasonality_mode": "multiplicative",
        "yearly_seasonality": true
    },
    "return_components": true
}

Response Format

{
    "forecast_dates": ["2023-02-01", ...],
    "forecast_values": [120.5, ...],
    "forecast_lower_bound": [115.2, ...],
    "forecast_upper_bound": [125.8, ...],
    "components": {
        "trend": [110.2, ...],
        "yearly": [10.3, ...],
        "weekly": [0.5, ...]
    }
}

Dependencies

• Requests
• FastAPI
• Numpy
• Prophet
• Pandas
• Pydantic
• Uvicorn