An Experimental Evaluation of Nearest Neighbour Time Series Classification

TL;DR

This paper evaluates the effectiveness of nearest neighbor classifiers with Euclidean and DTW distances for time series classification, questioning standard practices and exploring the impact of parameter tuning on performance.

Contribution

It provides an empirical comparison of 1-NN classifiers with Euclidean and DTW distances, analyzing the effects of parameter tuning and windowing on classification accuracy.

Findings

1-NN with Euclidean distance can be outperformed by other methods.

1-NN with DTW performs well when window size is optimized via cross validation.

Parameter tuning significantly influences the effectiveness of DTW-based classifiers.

Abstract

Data mining research into time series classification (TSC) has focussed on alternative distance measures for nearest neighbour classifiers. It is standard practice to use 1-NN with Euclidean or dynamic time warping (DTW) distance as a straw man for comparison. As part of a wider investigation into elastic distance measures for TSC~\cite{lines14elastic}, we perform a series of experiments to test whether this standard practice is valid. Specifically, we compare 1-NN classifiers with Euclidean and DTW distance to standard classifiers, examine whether the performance of 1-NN Euclidean approaches that of 1-NN DTW as the number of cases increases, assess whether there is any benefit of setting $k$ for $k$-NN through cross validation whether it is worth setting the warping path for DTW through cross validation and finally is it better to use a window or weighting for DTW. Based on experiments on 77 problems, we conclude that 1-NN with Euclidean distance is fairly easy to beat but 1-NN with DTW is not, if window size is set through cross validation.