Mean Absolute Error (MAE) is a popular metric used to evaluate the accuracy of a model's predictions, typically in regression problems. It measures the average absolute difference between the actual and predicted values of the target variable.
The formula for calculating MAE is:
MAE = (1/n) * Σ|AV(i) - PV(i)|
where n is the number of samples in the dataset, AV(i) is the actual value of the target variable for the i-th sample, and PV(i) is the predicted value of the target variable for the i-th sample.
Suppose we have a dataset that contains the weight and height of 10 individuals. We want to build a regression model that predicts the weight based on the height. Here is a sample dataset:
Sr.No. | Height(inches) | Actual Wt. | Pred-Wt. | (Actual-Predicted) wt. | absolute-diff |
---|---|---|---|---|---|
1 | 67 | 57 | 55 | 2 | 2 |
2 | 70 | 59 | 59 | 0 | 0 |
3 | 54 | 73 | 85 | -12 | 12 |
4 | 60 | 82 | 92 | -10 | 10 |
5 | 72 | 48 | 50 | -2 | 2 |
6 | 56 | 50 | 50 | 0 | 0 |
7 | 71 | 59 | 58 | 1 | 1 |
8 | 56 | 64 | 63 | 1 | 1 |
9 | 65 | 70 | 70 | 0 | 0 |
10 | 57 | 79 | 78 | 1 | 1 |
-- | -- | -- | -- | Sum = | 29 |
MAE = (Sum of absolute difference) / Count
MAE = 29 / 10
So, MAE = 2.9
This is the MAE for our regression model. It means that, on average, our model predicts the weight of an individual within +/- 2.9 Kgs. of the actual weight.
Advertisement
Advertisement