[76] - Machine Learning Applied to Fitness Data

Inspiration

As a bodybuilder and powerlifting, I love keeping track of my own stats in the gym. However, I also have a passion for Computer Science. My desire to combine these two interests has inspired the creation of this project.

What it does

First, the project trains a Support Vector Machine to draw classification boundaries between data points that correspond to the performance of a particular individual at a particular meet associated with three x variables: their max bench press, squat, and deadlift. The y variable is the participant's weight class.

After training, this SVM should be able to output a weight class given a lifters best bench press, squat, and deadlift.

How we built it

I used an OpenPowerlifting Dataset that can be accessed at https://openpowerlifting.gitlab.io/opl-csv/. The program itself is coded in python with the sklearn machine learning library.

Challenges we ran into

Preprocessing the data in the .csv document into a Numpy array, creating the SVM classifier, training that classifier, gaging the accuracy of the model, and using the classifier to make predictions on new input data were all challenges faced at the implementational level.

Accomplishments that we're proud of

The SVM correctly outputs a reasonable weight class given new input data.

What we learned

Machine learning has lots of everyday applications in surprisingly wide range of fields. Even something as seemingly unrelated as working out can be viewed through new perspectives by applying data analysis and machine learning techniques.

What's next for Machine Learning Applied to Fitness Data

If we had more data such as the height of participants, the model's accuracy can potentially be further improved. Moreover, the implementational insights gained from this project will be invaluable toward future machine learning projects I undertake.

Built With

• https://openpowerlifting.gitlab.io/opl-csv/bulk-csv.html
• https://www.openpowerlifting.org
• python