Lesson 2 Project

Neural Networks with Real Data

Complete this project to get experience with using real data in neural networks for regression prediction.

Extend the code from lesson 2 to use more than just 4 numerical inputs. Does this help with the network accuracy?

HINT:You'll need to change both the dataset and the neural network architecture to accommodate a larger input size.

Change the network architecture or learning rate using the network above to see how low you can get the loss of the network. Then save a copy of your best model, and load it in a new notebook.

Find another regression dataset on Kaggle, and build a network to make predictions with that net dataset.

Bonus: use your own dataset! Get one from work or school, or make one from online resources based on your hobbies. Make a CSV file with the dataset, and then load it up, and train a new neural network.

HINT:We've only learned about regression and numerical columns so far - so the input data and output data types must both be numbers.