Production-Ready PoC

BigQuery ML Time Series Forecasting

Complete ML pipeline using ARIMA_PLUS model, FastAPI backend, and real Bitcoin market data. Built in under 1 hour to demonstrate rapid prototyping capabilities.

View API Documentation → Interactive Visualization 7-Day Forecast

Key Features

Full-stack time series forecasting system with production-grade architecture.

BigQuery ML

ARIMA_PLUS model with automatic parameter tuning and seasonality detection.

Real Data

1,401 data points of actual Bitcoin market prices from CryptoCompare API.

REST API

FastAPI backend with async endpoints and automatic OpenAPI documentation.

Visualization

Interactive Plotly charts with historical data, forecasts, and confidence intervals.

Model Performance

Trained on real Bitcoin market data with detected weekly and yearly seasonality patterns.

1,401

Data Points

3.8Y

History

14D

Forecast Horizon

90%

Confidence

Technology Stack

Modern technologies with production-ready patterns and best practices.

BigQuery ML ARIMA_PLUS Python 3.12 FastAPI Pydantic Plotly Nginx SSL/HTTPS Docker-ready

API Endpoints

RESTful API with comprehensive documentation and interactive testing interface.

GET /api/forecast — Generate time series predictions
GET /api/model/evaluate — Model performance metrics
GET /api/model/explain — Forecast component breakdown
GET /api/visualize — Interactive Plotly visualization
GET /api/data/historical — Query historical time series