arXiv subjects

Nicholas Waltz

Publications and source records attributed to Nicholas Waltz.

2 recordsLinked to original sources

arXiv2024

Time Series Clustering Using DBSCAN

Economic policy and research rely on the correct evaluation of the billions of high-frequency data points that we collect every day. Consistent clustering algorithms, like DBSCAN, allow us to make sense of the data in a useful way. However, while there is a large literature on the consistency of various clustering algorithms for high-dimensional static clustering, the literature on multivariate time series clustering still largely relies on heuristics or restrictive assumptions. The aim of this paper is to prove a notion of consistency of DBSCAN for the task of clustering multivariate time series.

math.ST↗

arXiv2024

Grafting: Making Random Forests Consistent

Despite their performance and widespread use, little is known about the theory of Random Forests. A major unanswered question is whether, or when, the Random Forest algorithm is consistent. The literature explores various variants of the classic Random Forest algorithm to address this question and known short-comings of the method. This paper is a contribution to this literature. Specifically, the suitability of grafting consistent estimators onto a shallow CART is explored. It is shown that this approach has a consistency guarantee and performs well in empirical settings.

stat.ML↗