Individuals with Autism Spectrum Disorder (ASD) struggle with social interaction, communication, and repetitive behaviors, among other symptoms. Although total eradication is unlikely, early therapies may lessen the severity of the condition. In this study, we present a useful framework for comparing several Machine Learning (ML) approaches to ASD diagnosis at an early age. The suggested framework incorporates an effective prediction method into the overall design by using a variety of Feature Scaling (FS) techniques, including minmax scalar, principal component analysis, and intuitive visual representation analysis. Our suggested design using the IPD model shows that the overall verification of the features, various functional qualities reflects the adult autism spectrum disorder categorization in its entirety. In our IPD model, we use a chi-squared and probability density functional hybrid to do the mathematical analysis. The structure is meant to identify and categorize patients according to their observable characteristics. At the end of the day, we can see that the 95.9% accuracy is far higher than the Machine Learning method.

Improved Probabilistic Dense Model (IPD), Artificial Neural Network (ANN), Random Forest classifier, Support Vector Machine (SVM), Logistic Regression, Confusion Matrix (CM), Ensemble Approach (EA).