Implementing a Model to Predict… Hospital Readmission from Diabetes Diagnosis

This model utilizes several features of diabetes diagnosis to predict levels in patients’ hospital readmissions. The target of this paper is to analyze whether these diabetes diagnosis variables hold association in forecasting hospital readmission using decision tree classifiers, logistic regression, and discriminant analysis algorithms.

With machine learning, we hope to alleviate hospital operation costs by choosing which combination of measures of diabetes diagnosis will result in readmission. The length of the dataset (a decade) will provide us with enough observations to accurately predict readmission results. Ultimately, the goal of this analysis is to help companies within the healthcare industry to better allocate resources to strategically reduce the immense costs of hospital readmissions currently compromising company efficiency.

The dataset titled, Predicting Hospital Readmissions has been retrieved from the data science online platform Kaggle, containing historical data for 10 years of patient information. This data consists of both numerical and categorical variables, as well as calculated variables.