Exploring Clustering Techniques for Time Series Analysis: A Case Study on Moroccan Stock Market Data

Authors

Mahboub Sabah

Department of Modeling Applied to Economics and Management, University Hassan II Faculty of Judicious Economies and Social Sciences, Ain Sebaa, Casablanca, Morocco

National Institute of Statistics and Applied Economics, Rabat, Morocco

Raby Guerbaz

Department of Modeling Applied to Economics and Management, University Hassan II Faculty of Judicious Economies and Social Sciences, Ain Sebaa, Casablanca, Morocco

DOI

https://doi.org/10.62241/ijemd.22.0114.2339

Abstract

This study examines the effectiveness of clustering techniques for analyzing time series data in the context of the Moroccan stock market, focusing on 50 companies. The research utilizes Euclidean and Dynamic Time Warping (DTW) clustering methodologies to endeavor to discern latent trends and patterns in the oscillations of stock values across diverse firms. The strategy consists of multiple crucial steps. The time series data is first preprocessed to ensure its consistency and dependability. Then, it is determined how many clusters would be most effective in grouping the data. The DTW distance, which identifies nonlinear similarities in time series data, and the Euclidean distance, which emphasizes similarities in pricing patterns, are then used to cluster the data. The main conclusions of the study point to certain clustering tendencies in the Moroccan stock market dataset. While DTW clustering finds further details in nonlinear patterns that older approaches can miss, Euclidean clustering reveals groups of companies having similar price movements. These discoveries significantly expand on our understanding of the mechanics and composition of the Moroccan stock market. By using appropriate clustering algorithms, investors and financial analysts can gain valuable insights into market behavior and potentially more precisely pinpoint investment possibilities. The study highlights the need of selecting suitable analytical methods for time series data, especially in financial contexts where the ability to identify subtle patterns and trends is crucial for making well-informed decisions. Ultimately, our study advances academic knowledge and offers practical advice to professionals working in the financial and investing domains.

Pages

01-14

Citation