Median Absolute Error Ratio is the median of the absolute error ratio of the prediction for each piece of data included in the evaluation data. In this context, median refers to the value in the middle of a sequence of numbers in ascending order. In addition, the error ratio is the value obtained by dividing the error in numerical data by the correct value, and indicates the percentage of error with respect to the correct value.

It can be interpreted that approximately half of the data included in the evaluation data could be predicted with an error smaller than the median absolute error ratio.

Median values are aggregate values that are not easily affected by extreme values, so if you are interested in whether there are cases where the error is extremely large, you should also consider the mean absolute error and RMSE values.