From the course: Applied Machine Learning: Value Estimation
Join today to access over 25,500 courses taught by industry experts.
Evaluate a model: XGBoost
From the course: Applied Machine Learning: Value Estimation
Evaluate a model: XGBoost
- [Instructor] Okay, so let's evaluate our XGBoost Model. I'm going to look at the scores for both the training and the testing set, and this is something that we didn't see with linear regression. Notice here our R2 score for the training data is 0.99. Remember, this is a value that is ranging typically from zero to one. Our R2 score is very high on the training data. However, R2 score on the testing data is not as high. And what's happening here is that XGBoost is probably over fitting. What over fitting means is that your model is too complicated and it's pulling out signal from the noise in the data and makes it so it's not quite as able to generalize to data that it hasn't seen. Note that even though it is over fitting, it is giving better predictions on data that it hasn't seen than a linear regression model. So, what's my takeaway here? My takeaway is that it's probably possible to get even better performance on XG Boost by doing some tuning, getting that training score down so…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.