Influence over the Dimensionality Reduction and Clustering for Air Quality Measurements using PCA and SOM

The current trend in the industry is to analyze large data sets and apply data mining, machine learning techniques to identify a pattern. But the challenges with huge data sets are the high dimensions associated with it. Sometimes in data analytics applications, large amounts of data produce worse performance. Also, most of the data mining algorithms are implemented column wise and too many columns restrict the performance and make it slower. Therefore, dimensionality reduction is an important step in data analysis. Dimensionality reduction is a technique that converts high dimensional data into much lower dimension, such that maximum variance is explained within the first few dimensions. This paper focuses on multivariate statistical and artificial neural networks techniques for data reduction. Each method has a different rationale to preserve the relationship between input parameters during analysis. Principal Component Analysis which is a multivariate technique and Self Organising Map a neural network technique is presented in this paper. Also, a hierarchical clustering approach has been applied to the reduced data set. A case study of Air quality measurement has been considered to evaluate the performance of the proposed techniques.