A New Model for Autism Diagnosis Using Meta-Classifiers

Abstract

The illness that has drawn the greatest attention from scientists recently is autism, which is also the most well-known. People of various ages might be affected by this condition. All currently available datasets are of poor quality for data analysis. Base classifiers are the main focus of most related research. In this study, we proposed a thorough comparison that makes use of efficient preprocessing and ensemble approaches to enhance the diagnosis of the autistic condition. Outliers and missing data are resolved during the preprocessing step. The ensemble methods include bagging, stacking, boosting, and voting. Two well-known data mining technologies are used to assess comparisons. The findings collected to demonstrate that our work improves classification performance in terms of precision, recall, accuracy, and F1 compared to the base approaches. The criteria greatest values are attained at 100%.