model_evaluate

Description

Provides several evaluation metrics to assess the performance of the trained models. For regression, binary classification and multilabel classification different scoring metrics are presented to observe performance in many aspects.

Evaluation

This module is to considering performance results with many scoring metrics. Used metrics are

- Regression:
    - mean squared error
    - root mean sqeared error
    - Pearson correlation coefficient
    - Spearman's rank correlation coefficient
    - Threshold based classification Metrics
- Classification:
    - precision
    - recall
    - f1
    - f0.5
    - accuracy
    - Matthews correlation coefficient
    - AUROC
    - AUPRC

Explanation of Parameters

• model: Fitting to predict labels
• X: type = {list, numpy array}, feature matrix to introduce to model
• y: type = {list, numpy array}, corresponding label matrix
• preds: type = bool, If True return predictions and scores else only return scores
• isDeep: {bool}, default = False, If True, model is evaluated with torch.no_grad()
• learning_method: {"classif","reg"}, default = "classif", Learning task to get corresponding metrics

Usage

from profab.model_evaluate import evaluate_score
return_value = evaluate_score(model,
                            X,
                            y,
                            preds)

A use case:

score_test,f_test = evaluate_score(model,
                                X = X_test, 
                                y = y_test, 
                                preds = True)

Tabularizing the Metrics

To see scores in .csv files in an order, this function is proposed.

Explanation of Parameters

-- scores: type = {dict}, includes scores of sets (train, test) -- learning_method: {"classif","reg"}, default = "classif", to set values in order -- path: default = 'score_path.csv', destination where table will be saved. Format must be .csv

Usage

A use case:

from profab.model_evaluate import form_table

scores = {'test':score_test}

form_table(scores = scores)