 Similar to in classification we can use the logs to evaluate how model parameters change. So we say have our model dot make inspector training logs. I'm just going to immediately turn that into the data frame where we extract the number that should be quotes. The number of trees, and I'm once again going to do that sort of for loop where I say log dot num trees for every log in logs. And this time I want the root mean squared error. So I want log that valuation. That rooming squared error, or every log in logs. And so here we can see the number of trees, and we can see how the rooming squared error is. So if we plot this logs df with a line plot. Our x value is the number of trees. And our y value is the root mean squared error. We can see that as we increase the number of trees, our root mean squared error decreases. And eventually a plateaus. And so once again, similar to the classification analysis, we can use this to see if maybe we need more trees or less trees. And in this case, it seems like 200 might actually be a decent amount of trees. Perhaps an argument could be made that we should drop that down to 150 if we're trying to improve the speed of our model. But it certainly doesn't seem like we need any more trees in order to gain any more accuracy in our model.