Unit 7 Lesson 20: Troubleshooting Models

Hello! My students and I had a great discussion about potential problems with datasets based on the slide deck. However, when they used the datasets in Level 2 for Isaac’s Loneliness Predictor app, nobody was able to train an accurate model. The students noticed that there weren’t any hotspots in the data. Did we miss something? Are there combinations of features we didn’t try that would have led to an accurate model?


1 Like

Hi @msutton,

It looks like you did everything correctly. The conclusion is that all of the datasets are flawed in some way. Take a look at the lesson plan under the “Teaching Tips” tab and it goes through each dataset and its inherent flaw.


Oh! That makes sense. Of course. Thank you!