Estimated Time of Arrival Evaluator¶
We have implemented several evaluation loss functions so that different models under the same task can be compared under the same standard.
Evaluation Metrics¶
For the task of estimated time of arrival, this evaluator implements a series of evaluation indicators:
Evaluation Metrics |
Formula |
---|---|
MAE(Mean Absolute Error) |
\[MAE=\frac{1}{n}\sum_{i=1}^n|\hat{y_{i}}-y_i|\]
|
MSE(Mean Squared Error) |
\[MSE=\frac{1}{n}\sum_{i=1}^n(\hat{y_{i}}-y_i)^2\]
|
RMSE(Rooted Mean Squared Error) |
\[RMSE=\sqrt{\frac{1}{n}\sum_{i=1}^n(\hat{y_{i}}-y_i)^2}\]
|
MAPE(Mean Absolute Percent Error) |
\[MAPE=\frac{1}{n}\sum_{i=1}^n|\frac{\hat{y_{i}}-y_i}{y_i}|*100\%\]
|
R2(Coefficient of Determination) |
\[R^2=1-\frac{\sum_{i=1}^n(y_i-\hat{y_i})^2}{\sum_{i=1}^n(y_i-\bar{y})^2}\]
|
EVAR(Explained variance score) |
\[EVAR =1-\frac{Var(y_i-\hat{y_i})}{Var(y_i)}\]
|
The ground-truth value is \(y=\{y_1,y_2,...,y_n\}\), the prediction value is \(\hat{y} = \{\hat{y_1}, \hat{y_2}, ..., \hat{y_n}\}\),\(n\)is the number of samples, the mean value is \(\bar{y}=\frac{1}{n}\sum_{i=1}^ny_i\), the variance is \(Var(y_i)=\frac{1}{n}\sum_{i=1}^n(y_{i}-\bar{y})^2\).
Evaluation Settings¶
The following are parameters involved in the evaluator:
Location: libcity/config/evaluator/ETAEvaluator.json
metrics
: Array of evaluation metrics,allowed_metrics
in evaluator class indicates the type of metrics that the task can accept, andmetrics
cannot exceed this range.