Benchmarking Time Series Forecasting Models: From Statistical Techniques to Foundation Models in Real-World Applications

TL;DR

This paper benchmarks various time series forecasting models, including statistical, ML, deep learning, and foundation models, on real-world restaurant sales data, highlighting the strengths of ML meta-models and foundation models like Chronos and TimesFM.

Contribution

It introduces a comprehensive evaluation of foundation models for large-scale, real-world time series forecasting and demonstrates their competitive performance with minimal feature engineering.

Findings

ML-based meta-models outperform traditional statistical methods.

Foundation models like Chronos and TimesFM show promising zero-shot capabilities.

Hybrid PySpark-Pandas approach enables scalable deployment.

Abstract

Time series forecasting is essential for operational intelligence in the hospitality industry, and particularly challenging in large-scale, distributed systems. This study evaluates the performance of statistical, machine learning (ML), deep learning, and foundation models in forecasting hourly sales over a 14-day horizon using real-world data from a network of thousands of restaurants across Germany. The forecasting solution includes features such as weather conditions, calendar events, and time-of-day patterns. Results demonstrate the strong performance of ML-based meta-models and highlight the emerging potential of foundation models like Chronos and TimesFM, which deliver competitive performance with minimal feature engineering, leveraging only the pre-trained model (zero-shot inference). Additionally, a hybrid PySpark-Pandas approach proves to be a robust solution for achieving horizontal scalability in large-scale deployments.