Dynamic Time Warping

Dynamic Time Warping (DTW) calculates the similarity between two temporal sequences that vary in speed or timing.

DTW is a robust algorithm used to measure the distance between time-series data by finding an optimal alignment through non-linear stretching. Unlike Euclidean distance, which compares points at fixed timestamps, DTW allows for shifts in the time axis to match similar shapes (patterns) across different durations. It is a standard in speech recognition for matching spoken words to templates and in finance for identifying recurring market trends. By utilizing dynamic programming to minimize a cumulative cost matrix, DTW effectively handles signals with varying sampling rates or phase shifts.

https://databricks.com/blog/2019/04/30/understanding-dynamic-time-warping.html