A Stack Based Ensemble Learning Method for Diagnosing Autism Spectrum Disorder
Autism Spectrum Disorder (ASD) is a spectrum of neurodevelopment impairments that affects the nervous system and also affects individual's over all cognitive, emotional, social, and physical health. This disorder can be observed at early stages of life. In developmental stages, i.e., within the first two years after birth its symptoms are usually shown. The only approach for diagnosing ASD is using through Clinical standardized tests. This not only necessitates a longer diagnosis time, but it also results in a significant rise in medical costs. To improve the precision and time required for diagnosis, single machine learning techniques are being used now days but these are not sufficient for effective prediction of ASD. In this study, introduced an ASD prediction model for children dataset using Stack ensemble learning approach and which complements to the conventional single learning methods. In this study ASD predictive analysis is done in three stages. At first stage, applied feature selection method i.e., Principal Component Analysis (PCA) to retrieve optimal features of children data set. Later, in second stage, ASD is diagnosis with predictive model is designed using various base classification techniques which includes Random Forest Classifier (RFC), Naïve Bayes (NB), and Logistic Regression (LR) to children dataset. At the last stage, to improve the performance of ASD prediction model, applied stack ensemble learning method with mentioned base learners. The results shows that proposed ensemble based predictive model is best for ASD diagnosis for the children dataset at early stages than the base Machine Learning methods in terms of accuracy, precision, Recall and F1- Score.