Milestone Project 3: Function That Retrains Model After Making Prediction

[wanderduck]

Could someone walk me through or give me some hints at what exactly would need to be done to make a function that not only makes a future prediction, but also retrains the model with the prediction data? For whatever reason I'm having trouble wrapping my brain around the problem and how to go about solving it.

For reference this is a suggested side project in the model 9 section of the milestone project 3 where you are predicting bitcoin's price.

[mrdbourke]

Hey @wanderduck,

In pseudo-code this might like something like:

have trained model
make predictions on data
retrain model with data.append(predictions)
make predictions on updated data with retrained model
retrain model with data.append(predictions)
make predictions on updated data with retrained model
... (keep going until finished desired number of steps)

You could have a function doing each of these steps.

E.g. one for the original trained model, one for updating the dataset and one for retraining a model with another dataset. Then step through each of the functions for as many steps as you need.