In this notebook, we will explore using Bayesian Logistic Regression in order to predict whether or not a patient has diabetes. The task is supervised - we are given labeled training data to fit our model - and it is a classification task because the lables are binary. Typically this is a task that is approached with supervised machine learning techniques such as logistic regression, support vector machines, or tree based methods. We will take a slightly different approach and use a Bayesian Framework to fit a logistic regression model and then intrepret the resulting model parameters
Predict whether or not the patient has diabetes. To accomplish this task, we will use a set of real-world data collected on females 21 years of age and over collected by a national health institution in the United States.
Overall, this project provided a nice introduction to the Bayesian approach to linear modeling. While not applicable in all situations, in problems with limited data, Bayesian methods allow us to create an interpretable model with good performance while also demonstrating our uncertainty about the model. Making accurate predictions is useful by itself, but it is even more powerful when we can explain the predictions of our model and use the results to improve real-world outcomes.