Autism-Diagnosis-with-Supervised-Machine-Learning-and-Neural-Networks-using-Resampling

Organization

In the "Autism Prediction" Folder you will find "Cleaning", "Data", and "Machine Learning Models" folders as well as our project proposal and presentation pdf.

Data

This folder contains the original dataset which was procured from Kaggle. It also contains the cleaned dataset, the dataset which only includes ASQ scores and the weighted score, and the dataset which only includes the raw ASQ scores. This folder also contains two excel files which describe the accuracy scores and were used for visual generation in Tableau.

Cleaning

This folder contains the python code for the cleaning of our original train dataset into the Full Set, ASQ, and ASQ unweighted datasets.

Analyses

This folder contains the three folders which focus on the distinct "Full Set", "ASQ", and "Unweighted ASQ" datasets. Each folder contains the code to construct three neural network models and two linear regression models using the respective dataset.

Presentation

The Predicting Autism with Machine Learning walks through our project, providing explanations about what autism is, why machine learning models would be useful, and how we conducted our cleaning and analysis.

Conclusion

We were able to generate machine learning models that could predict autism with 90% accuracy. This could be an invaluable tool to the autistic community including parents, teachers and autistic individuals.