Dataiku's Solution to SPHERE's Activity Recognition Challenge

TL;DR

This paper describes Dataiku's winning solution to the SPHERE activity recognition challenge, utilizing advanced data preprocessing, synthetic data generation, feature engineering, stacking weak learners with XGBoost, and post-processing for smoothing predictions.

Contribution

The paper introduces a comprehensive pipeline combining synthetic data generation, feature engineering, stacking models, and post-processing for activity recognition from sensor data.

Findings

Achieved second place in the challenge

Effective synthetic data generation matching test set statistics

Improved prediction stability through post-processing

Abstract

Our team won the second prize of the Safe Aging with SPHERE Challenge organized by SPHERE, in conjunction with ECML-PKDD and Driven Data. The goal of the competition was to recognize activities performed by humans, using sensor data. This paper presents our solution. It is based on a rich pre-processing and state of the art machine learning methods. From the raw train data, we generate a synthetic train set with the same statistical characteristics as the test set. We then perform feature engineering. The machine learning modeling part is based on stacking weak learners through a grid searched XGBoost algorithm. Finally, we use post-processing to smooth our predictions over time.