This project investigated large-scale misrepresentation of intein-containing protein records in the National Center for Biotechnology Information (NCBI) databse using legacy intein datasets from prior peer reviewed research.
Download Write Up Download PresentationA UCI dataset with 3 clinical and 20 molecular features was analyzed to classify glioma grades. Logistic Regression with 10-fold cross-validation yielded: 87.3% accuracy, 80.2% precision, 92.9% recall.
View on Kaggle Read the Write-upA Kaggle dataset including variables like BMI, Glucose, and HbA1c-level was cleaned and modeled using Linear Regression. Though the data source was unverified, it was a fast-paced modeling experiment.
View on KaggleBased on reprocessed UCI Cleveland data, this project compares Logistic, Linear, Decision Tree, and Random Forest models. Variants of the dataset were explored for predictive accuracy.
Original Dataset View on Kaggle Download PDF