When I heard about LendingClub's public loan dataset, containing all loans issued by the company from it's launch in 2007 until the end of 2018, I figured that'd be the perfect opportunity to build a predictive model.
I wrote an article detailing my entire process, which you can read on my blog, Towards Data Science, Hacker Noon, or DEV. If you'd like to follow along in your own Jupyter Notebook, you can go ahead and fork mine on Kaggle or here on GitHub.
After building the model itself, I built an API to serve its predictions, using Flask, TensorFlow/Keras, pandas, and scikit-learn. You can interact with the API by either visiting its demonstrational front end or sending a GET request directly to https://tywmick.pythonanywhere.com/api/predict. The front end site includes a form where you can fill in all the parameters for the API request, and there are a couple of buttons at the top that let you fill the form with typical examples from the dataset (since there are a lot of fields to fill in).
I later wrote a couple of follow-up posts expanding the project:
Please enjoy, and let me know if you have any questions!