Model tester¶
-
class
pyfume.Tester.
SugenoFISTester
(model, test_data, variable_names, golden_standard=None, list_of_outputs=['OUTPUT'])¶ Bases:
object
Creates a new Tester object to be able to calculate performance metrics of the fuzzy model.
- Parameters
model – The model for which the performance metrics should be calculated
test_data – The data to be used to compute the performance metrics
variable_names – A list of the variables names of the test data (which should correspond with the variable names used in the model).
golden_standard – The ‘True’ labels of the test data. If not provided, the only predictions labels can be generated, but the error will not be calculated (default = None).
list_of_outputs – List of the output names (which should correspond with the output names used in the model) (default: OUTPUT).
-
calculate_MAE
()¶ Calculates the Mean Absolute Error of the model given the test data.
- Returns
The Mean Absolute Error of the fuzzy model.
-
calculate_MAPE
()¶ Calculates the Mean Absolute Percentage Error of the model given the test data.
- Returns
The Mean Absolute Percentage Error of the fuzzy model.
-
calculate_MSE
()¶ Calculates the Mean Squared Error of the model given the test data.
- Returns
The Mean Squared Error of the fuzzy model.
-
calculate_RMSE
()¶ Calculates the Root Mean Squared Error of the model given the test data.
- Returns
The Root Mean Squared Error of the fuzzy model.
-
calculate_performance
(metric='MAE')¶ Calculates the performance of the model given the test data.
- Args:
metric: The performance metric to be used to evaluate the model. Choose from: Mean Absolute Error (‘MAE’), Mean Squared Error (‘MSE’), Root Mean Squared Error (‘RMSE’), Mean Absolute Percentage Error (‘MAPE’).
- Returns
The performance as expressed by the chosen performance metric.
-
predict
()¶ Calculates the predictions labels of the test data using the fuzzy model.
- Returns
- Tuple containing (result, error)
result: Prediction labels.
error: The difference between the prediction label and the ‘true’ label.