Feature Programming for Multivariate Time Series Prediction

TL;DR

This paper presents a programmable feature engineering framework for multivariate time series prediction, leveraging a novel spin-gas Ising model to generate predictive features and improve modeling of noisy data.

Contribution

It introduces a new feature programming framework that automates feature generation for multivariate time series using a spin-gas Ising model, incorporating user inductive bias.

Findings

Effective on synthetic and real-world noisy datasets

Automates large-scale feature generation

Improves prediction accuracy

Abstract

We introduce the concept of programmable feature engineering for time series modeling and propose a feature programming framework. This framework generates large amounts of predictive features for noisy multivariate time series while allowing users to incorporate their inductive bias with minimal effort. The key motivation of our framework is to view any multivariate time series as a cumulative sum of fine-grained trajectory increments, with each increment governed by a novel spin-gas dynamical Ising model. This fine-grained perspective motivates the development of a parsimonious set of operators that summarize multivariate time series in an abstract fashion, serving as the foundation for large-scale automated feature engineering. Numerically, we validate the efficacy of our method on several synthetic and real-world noisy time series datasets.