0

I used MLR package to create a model in R to predict the price using distance.( Linear regression model). on doing the performance metric check - I get the below output : mse 0.01985664 0. mae 0.11441877 How do I interpret it ? I know that its good to have a small absolute error, but can I say that my model is statistically significant and good for use ? whats the right wording.?

analyst045
  • 628
  • 7
  • 21
  • To tell whether something is statistically significant you need an experimental design generally, and then something like an F-test. mse will just tell you how your model performed against the data you gave it. You can try to implement this model on new data and see your performance, that's the general way I believe. – Alexis Drakopoulos Oct 20 '18 at 13:22
  • This sort of question is better suited for [Cross Validated](https://stats.stackexchange.com/), and will likely be migrated there. In the meantime, I'll address some of the content of [Alexis Drakopoulos](https://stackoverflow.com/users/8852013/alexis-drakopoulos)'s comment. (1) You do not need an experimental design to test for statistical significance, though experimental designs are often useful to address causality. (2) Alexis is right that to test for overall model significance is an F test; it will tell you whether your model's fit is significantly different from an intercept only model. – duckmayr Oct 20 '18 at 13:41
  • [Here](http://blog.minitab.com/blog/adventures-in-statistics-2/what-is-the-f-test-of-overall-significance-in-regression-analysis) is a fairly accessible blog post on the F test for overall model significance, though I would strongly suggest to you to do more studying before really trying to interpret the results of your model. – duckmayr Oct 20 '18 at 13:43
  • Thanks Alexis and duckmayr. – analyst045 Oct 20 '18 at 23:38
  • You can also calculate the Rsquared (rsq in mlr), that gives you a scaled version of the mse that lies between 0 and 1 (1 is the best value). This is easier to interpret and shows how well your model performs compared to a simple mean. – PhilippPro Oct 23 '18 at 09:06
  • thanks Philipp ...will give it a try – analyst045 Oct 24 '18 at 12:25

0 Answers0