Researched by Liz Jones, Melissa Keller, & Vallie Tracy
Finished Website
A snapshot of our website's homepage
In this repo, we explore using machine learning to detect autism in adults. We inititally found the dataset on the UCI Machine Learning Repository website. After doing some digging, we found the author's homepage which contains the most current dataset. You can find both the csv and original publication by navigating to our Resources folder.
The data was compiled in 2018 by Dr. Fadi Thabtah as he and his colleagues explored a new approach to machine learning for autism, called Rules-Machine Learning, based on the Covering approach. Thabtah and colleagues explored datasets grouped by toddler, child, adolescent, and adult. In this repo, we focus on the adult dataset.
In our analysis, we looked at three different classifier approaches: Logistic Regression, K-Nearest Neighbors, and Random Forest. We also began to delve into K-means. We analyzed the dataset itself as well. To read further about our observations on both, please visit our website, whose link is found at the top of this Read Me.
PROGRAMMING LANGUAGES:
- Python
PACKAGES/LIBRARIES USED:
- Sklearn
- Joblib
- Pandas
- NumPy
- Seaborn
- Matplotlib
ML OPTIMIZATION TOOLS:
- GridSearchCV for hyperparameter tuning
- MinMaxScaler for preprocessing
- Feature Selection: ExtraTreesClassifier, RandomForestClassifier, LogisticRegression.coef_
VISUAL INTERFACE:
- HTML
- Bootstrap