Machine Learning

  Precision and Recall Scikit-Learn provides several functions to compute classifier metrics , including precision and recall : from sklearn.metrics import precision_score , recall_score precision_score ( y_train_5 , y_train_pred ) # == 4096 / (4096 + 1522) 0.7290850836596654 recall_score ( y_train_5 , y_train_pred ) # == 4096 / (4096 + 1325) 0.7555801512636044 ·        Now your 5-detector does not look as shiny as it did when you looked at its accuracy . ·        When it claims an image represents a 5 , it is correct only 72.9% of the time. ·        Moreover, it only detects 75.6% of the 5 s. •         It is often convenient to combine ü   precision and ü   recall into a single metric called the F1 score , ü   in particular if you need a simple way to compare two classifiers . •         The F1 score is the harmonic mean of precision and recall . F1Score     •         Whereas the regular mean treats all values equally , ü   the ha

  Confusion Matrix ·        A much better way to evaluate the performance of a classifier is to look at the confusion matrix .   ·        The general idea is to ·        count the number of times instances of class A are classified as class B. ·        For example, ü   to know the number of times the classifier confused images of 5s with 3s , ü   you would look in ·        the fifth row and ·        third column of the confusion matrix . ü   To compute the confusion matrix , •         you first need to have a set of predictions so that they can be compared to the actual targets . •         You could make predictions on the test set. •         Remember that you want to use the test set only at the very end of your project , once you have a classifier that you are ready to launch . ü   Instead, you can use the cross_val_predict() function: from sklearn.model_selection import cross_val_predict y_train_pred = cross_val_predict ( sgd_clf , X_t

 Machine Learning - Classification - MCQs 1.      By default, SGD classifier follows this strategy for multi-class classification: Top of Form A)    OvO strategy B)     OvA strategy C)     Both D)    None Ans: B 2.      SGD Classifiers and Linear Classifiers are strictly A)    Binary Classifier B)     Multiclass classifier C)     Both D)    None      Ans: A 3.      The greater the value for ROC AUC, better the model: Top of Form A)    True B)     False Bottom of Form Ans: A 4.      The maximum value of the ROC AUC is Top of Form A)     0.8 B)      0.9  C)      1 D)     0.7 Bottom of Form Ans: C 5.      Recall can be increased by increasing the decision threshold. True or False Top of Form ? A)     False B)      True Ans: A 6.      Precision can be increased by increasing the decision threshold. True or False? A)    True B)     False Ans: A   7.      Which of these is a good measure to decide which threshold to use?