Statistics Colloquium by Eamon Gara Grady ’23

Sparse Principal Component Analysis and Its Application to Soccer Player Performance Data by Eamon Gara Grady ’23, Statistics Colloquium, Wednesday, May 3, 1:10 – 1:50 pm, North Science Building 114, Wachenheim.

Abstract:  Sparse principal component analysis (SPCA) is a dimension reduction technique often applied when working with high-dimensional data.  It works by forming orthogonal predictors from linear combinations of features in a dataset as a regular PCA does.  It introduces sparsity in the loadings to make the  interpretations of principal components easier than that from a regular PCA. In this talk, I will first introduce SPCA, and then discuss an application of SPCA to reduce the dimensionality of soccer players’ attributes derived from their EA Sports FIFA ratings.  Then I will discuss how the results of SPCA were used to group like players together with the k-means clustering method.