Application of Dimensionality Reduction techniques in Real time Dataset

Data pre-processing in data mining refers to transforming the raw data into understandable format for further analysis. The real time data is incomplete, robust and unorganised which should be cleaned and transformed to make it efficient for preprocessing. In this paper, we have discussed about three dimensionality reduction techniques namely Principal Component Analysis (PCA), Singular Value decomposition and Learning Vector Quantization applied to solar irradiance dataset. The dataset consists of temperature, solar irradiance, and humidity data for 25 years for selected eight south Indian cities. The dimensionality reduction was done by applying the above mentioned three algorithms and their efficiency was evaluated, to find the best suiting algorithm to apply for the dataset. Index Terms Data Mining, Dimensionality Reduction, Learning Vector Quantization, Principal Component Analysis (PCA), Singular Value Decomposition.