Metrics¶
The metric of an experiment is the function that will be used to assess for the efficiency of its models. The metrics you can choose depends on the type of experiment you are training.
-
class
previsionio.metrics.
Classification
¶ Metrics for classification projects Available metrics in prevision: auc, log_loss, error_rate_binary
-
AUC
= 'auc'¶ Area Under ROC Curve
-
AUCPR
= 'aucpr'¶ precision recall area under the curve score
-
F05
= 'F05'¶ F05 Score
-
F1
= 'F1'¶ Balanced F-score
-
F2
= 'F2'¶ F2 Score
-
F3
= 'F3'¶ F3 Score
-
F4
= 'F4'¶ F4 Score
-
Lift01
= 'lift_at_0.1'¶ lift at ratio 0.1
-
Lift02
= 'lift_at_0.2'¶ lift at ratio 0.2
-
Lift03
= 'lift_at_0.3'¶ lift at ratio 0.3
-
Lift04
= 'lift_at_0.4'¶ lift at ratio 0.4
-
Lift05
= 'lift_at_0.5'¶ lift at ratio 0.5
-
Lift06
= 'lift_at_0.6'¶ lift at ratio 0.6
-
Lift07
= 'lift_at_0.7'¶ lift at ratio 0.7
-
Lift08
= 'lift_at_0.8'¶ lift at ratio 0.8
-
Lift09
= 'lift_at_0.9'¶ lift at ratio 0.9
-
MCC
= 'mcc'¶ Matthews correlation coefficient
-
accuracy
= 'accuracy'¶ Accuracy
-
error_rate
= 'error_rate_binary'¶ Error rate
-
gini
= 'gini'¶ Gini score
-
log_loss
= 'log_loss'¶ Logarithmic Loss
-
-
class
previsionio.metrics.
MultiClassification
¶ Metrics for multiclassification projects
-
AUC
= 'auc'¶ Area Under ROC Curve
-
MAP10
= 'map_at_10'¶ qmean average precision @10
-
MAP3
= 'map_at_3'¶ qmean average precision @3
-
MAP5
= 'map_at_5'¶ qmean average precision @5
-
accuracy
= 'accuracy'¶ accuracy
-
error_rate
= 'error_rate_multi'¶ Multi-class Error rate
-
log_loss
= 'log_loss'¶ Logarithmic Loss
-
macroF1
= 'macroF1'¶ balanced F-score
-
qkappa
= 'qkappa'¶ quadratic weighted kappa
-
-
class
previsionio.metrics.
Regression
¶ Metrics for regression projects Available metrics in prevision: rmse, mape, rmsle, mse, mae
-
MAE
= 'mae'¶ Mean Average Error
-
MAPE
= 'mape'¶ Mean Average Percentage Error
-
MER
= 'mer'¶ Median Absolute Error
-
MSE
= 'mse'¶ Mean squared Error
-
R2
= 'R2'¶ R2 Error
-
RMSE
= 'rmse'¶ Root Mean Squared Error
-
RMSLE
= 'rmsle'¶ Root Mean Squared Logarithmic Error
-
RMSPE
= 'rmspe'¶ Root Mean Squared Percentage Error
-
SMAPE
= 'smape'¶ Symmetric Mean Absolute Percentage Error
-