Following PR #47, one should provide visualization of the data. I rephrased my previous suggestions to below:
Add either in one or both exercise and solution, a plot of the data before running the model, preferable after one has loaded the data. It is useful to list out a few data points (in pandas, that's the head method) and/or plot the graph. In that way, the learners understand how/why we are using linear regression, instead of just blindly running the model
After training (20,000 epochs!), plot the data with the model predictions from linear regression? Optionally/bonus, if one could plot a few predictions, say one from after 1000 epochs, one from 10,000 epochs and one from 20,000 epochs. This illustrates how the fitting get better and better (if it doesn't, then we should have stopped much earlier in our training process).
Following PR #47, one should provide visualization of the data. I rephrased my previous suggestions to below:
Add either in one or both exercise and solution, a plot of the data before running the model, preferable after one has loaded the data. It is useful to list out a few data points (in pandas, that's the head method) and/or plot the graph. In that way, the learners understand how/why we are using linear regression, instead of just blindly running the model
After training (20,000 epochs!), plot the data with the model predictions from linear regression? Optionally/bonus, if one could plot a few predictions, say one from after 1000 epochs, one from 10,000 epochs and one from 20,000 epochs. This illustrates how the fitting get better and better (if it doesn't, then we should have stopped much earlier in our training process).