Diabetes Prediction

Inspiration

Based on existing information regarding an individual’s health conditions and lifestyle habits, can we accurately diagnose whether they have diabetes or pre-diabetes?

What it does

Predicting if a person has diabetes or pre-diabetes.

How we built it

We explored the data and tried various machine learning models from sklearn such as SVMs, Random Forests and Decision Trees. We also implemented a fully connected neural network with Tensorflow/Keras having 3 hidden layers.

Challenges we ran into

Selecting the right features with high accuracy predictions.

Accomplishments that we're proud of

Relatively high accuracy of predicting diabetes (or pre-diabetes).

What we learned

We explored many different Machine Learning architectures of predicting medical results, which is very important for diagnosis in practice.

What's next for Diabetes Prediction

Exploring the other unbalanced data sets and improving the accuracy.

Built With

• decisiontree
• deep-learning
• jupyter-notebook
• machine-learning
• python
• randomforest
• svm