Actual vs Predicted

General
1

Dear Community,

I have a quick question:
I used a multiple regression to predict the price of real-estate. The population has been divided into a training and testset. The regression model has predicted the prices for the testset and I also know the differences between the actual prices and the predicted prices.

Now I want to know if the difference between actual prices and predicted prices is significant. I want to do this with a t-test and made the differences absolute and performed a two-tailed t-test:
t.test(verschillen.test$verschil, mu=0, alt = "two.sided")

H0: the difference between predicted and actual is 0

But i'm kind of doubting this approach.
Is it possible to test this with a t-test ?
and should I make the difference absolute? I did this because otherwise large negative differences and positives differences will cancel each other out?

Thank you in advance! :slight_smile:

With kind regards

2

Generally, different tests of accuracy are used, such as RMSE, MAE, MAPE, etc. See Hyndman

3

Thank you technocrat, I will check the MSE.

1 Like
4

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.