Redução de Dimensionalidade Via Análise de Componentes Principais de Variáveis Inerentes à Geração de Energia Hidrelétrica

Abstract

The increased amount of monitored data allows more complex analysis and optimization techniques. As a result of the high dimensional databases, the computational cost has become a critical point. This paper proposes the application of the Principal Component Analysis (P.C.A.) to reduce high dimensional databases, applied to a hydroelectrical generation dataset, and evaluate its impacts on supervised machine learning techniques prediction. The findings of this study showed that by applying the P.C.A. the reduced dataset was able to preserve most of the data variability of original data and achieved a similar performance rate on machine learning techniques, compared to the non-reduced dataset.