Many today are quite cautious about their own weight, constantly trying to understand whether their body composition fits the standard. Most of the time, those who want to get an in-depth and highly accurate understanding of their body fat percentage must spend a pretty penny. I found this to be a great opportunity to train a machine learning model based on the recorded data of 252 participants containing their body measurements and their assessed body fat percentages through hydrodensitometry better known as underwater weighing. This would allow for a great opportunity to predict body fat percentage without the hassle of underwater weighing and time taken and simply with one’s own body measurements.
Although the sample size might seem too small, the model surprisingly makes very accurate predictions when entering proper measurements. The fully deployed application shows different visualizations pertaining to the user’s weight class, with an option to compare the user’s weight to all of the participants as a whole. If someone does not know their body measurements or have the time to actually get them they can still make use of the app by testing a sample of the data to get a better insight into where most participants lie.
Rather than using Flask, Streamlit gives Python developers a great opportunity to publish and deploy their code on their website. Whether it’s a machine learning model or a simple dashboard that reflects data pulled from an API, it seamlessly makes life easier. Their framework gave me a great opportunity to show the model off to non-technical users.
For more information on the entirety of the project including algorithms used and prediction accuracy, make sure to check out the links below which include the full source code on Github, end-to-end deployment, exploratory data analysis, and a Tableau dashboard of the dataset.
Github: Source Code
Deployment: Streamlit Webapp
Tableau Dashboard: Tableau Public or Download