Research on Product Feature Mining Based on Big Data and K-Means Clustering Algorithm

The aim of this paper is to analyse and study the feature relationships of the targets through various clustering algorithms. The study first cleans and preprocesses the data and selects average monthly transaction volume, average monthly transaction value, unit price, transaction growth rate and volatility as key indicators. Then, K-Means, DBSCAN and spectral clustering algorithms are used for clustering analysis, and K-Means is selected as the optimal model by comparing the model performance. The clustering results show that the five products can be classified into three categories. In addition, this study used multiple linear regression models to analyse the effects of month, year and unit price on transactions. By combining multiple clustering algorithms and regression analyses, this study provides insights into understanding the similarities between product brands and their transaction characteristics, which provides an important reference value for market analyses and decision-making in this product industry.