Music Evaluation Based on Principal Component Analysis and Pearson Correlation Coefficient
Download as PDF
DOI: 10.23977/iset2021.007
Author(s)
Yike Xu, Zujun Hu
Corresponding Author
Yike Xu
ABSTRACT
Music and society are mutually constitutive. Initially, a preliminary data cleaning was conducted. And in order to understand the influence relations between various artists, we develop the directed influence network model. Meanwhile, ArticleRank was employed as a quantitative indicator to capture musical influence. With mounting quantities of data, we created subnetworks to explore the subsets of musical influence in more detail. Subsequently, at the request of capture the effective information accurately, we reduced the dimension of musical data by utilizing Random Forest and PCA analysis methods. To address the problem of comparing the similarity of music, Pearson Correlation Coefficient (PCC) was introduced the similarity measurement model was successfully constructed. Founded on what we have done, we accomplished the similarity analysis, in addition, we clustered genres by k-means algorithms and set up the genre development and comparison model to represent the similarities and influences between and within genres.
KEYWORDS
PCA, Pearson Correlation Coefficient (PCC), k-means algorithms