v3551g / blood-pressure-analysis-and-prediction

Data analysis and machine learning model creation for predicting blood pressure from the National Health and Nutrition Examination Survey data set.

Blood Pressure Analysis and Prediction

Josh McComack

Overview

In this Jupyter Notebook I investigate the feasibility of predicting systolic and diastolic blood pressure based on the National Health and Nutrition Examination Survey (NHANES) data set from 2013-2014 found on kaggle.com. The primary questions I wish to answer are:

1. Which variables from the survey are most predictive empirically and do they correspond to the what mainstream literature identifies as key factors in blood pressure levels?
2. Comparing scikit-learn’s SGDRegressor, MultiTaskLasso, and RandomForestRegressor, which regression model offers the best predictions on this data set?
3. Does the best model perform well enough to serve as a possible supplementary or alternative way of “measuring” blood pressure?

Install

download and install the Anaconda package manager.

Launch

Open a terminal in the root directory of this project and enter:

jupyter notebook