Impact of ranked ordered feature list (ROFL) on classification with visual data mining techniques

摘要

Classification of data is used in data analysis to group various instances in appropriate classes to enhance readability of data and study its characteristics easily. The main aim of every classification problem is the enhancement of classification accuracy. Ranked feature ordering helps in improving the classification accuracy by removing the least dominant features. Classification model uses only important features and eliminate least dominant, results obtained later can be better understood by plotting them in parallel plot using visual representation. Visual representation of classifier results helps in better comprehension and interpretation of results. Parallel plot is a one type of coordinate plot, that have ability to show any number of variables in one plane, the plane is either two dimensional plane or it is three dimensional plane. Parallel plot also shows relationships between variables. Proposed article used Electroencephalography (EEG) eye state classifier data for this account.