Statistical and unsupervised mls analysis on parkinson’s disease data set acquired from A.P. India

Abstract

In recent years, the voice analysis is the important work for identifying the neurological diseases like Parkinson’s disease (PD). PD is the subsequent general neurodegenerative disorder after Alzheimer’s lacking of dopamine in mid brain. In most people, symptoms appear at the age of 50 years or over. In this research, one thousand two hundred vowel-sounded (/.a/.e/.i/.o/.u) voice records are collected from A.P., India for analyzing people with PD from that of healthy people. The records constitute the data of 40 PD patients and 36 non-PD people who are having their age between 50 and 85. Those voice recordings are processed and relevant features or characters are extracted. Here, the data set contains features of both people with PD and healthy to distinguish performance. In this, we analyzed the PD dataset with statistical and unsupervised machine learning analysis. The efficient clustering k-means algorithm represents the Centroids of each attribute of the PD voice data set in two clusters (cluster 0, cluster 1). Another used unsupervised ML algorithm, hierarchal clustering clusters the data set in row wise (attribute wise) as well column wise(data wise) and analyze the projections of attributes and their rankings with using PCA(Principle component analysis). Parkinson’s disease (PD), Unsupervised Machine Learning, Voice, PCA.