Logistic principle component analysis (L-PCA) for feature selection in classification

In Data Mining, especially when datasets are complex with many features existed. Data preparation and analysis may be necessary in implement effective data mining model, especially in classification of high dimensional data where not all features are important in the mining process. Hence, feature selection is an essential in data preprocessing. A well known technique in selecting useful attributes is Principle Component Analysis (PCA). The technique assigns a value to each component (feature), so order of importance among components can be apparent. This work presents a new feature selection method. It improves the effectiveness of PCA by means of utilizing the Logistic Function to become Logistic Principle Component Analysis (L-PCA). PCA and L-PCA are compared by means of classifying ten public domain datasets. The result reveals that LPCA is superior to PCA and able to select crucial features more efficiently.