Exam AWS Certified Machine Learning - Specialty topic 1 question 82 discussion

Answer B.

B IS NOT CORRECT! Return the probability. Not the 1 or 0. D IS THE CORRECT ANSWER.

Streaming classification: is the process of organizing and categorizing large amounts of data that are continuously flowing. This data can include medical records, banking transactions, and internet records Binary Classification: Logistic Regression Multiclass Classification: Softmax regression Regression Classification is a made-up term

I always see that the community voting is more appropriate and the moderator answer looks out to be on wrong side. I see this for almost in 1 out of 5 questions. Which answer should we consider here as right one ??

Its definitely a classification problem, and between Binary and Streaming classification. Binary classification makes more sence

Binary classification

B, easy.

B, obviously! from sklearn.linear_model import LogisticRegression log_reg = LogisticRegression() log_reg.fit(X_train, y_train) log_reg.predict_proba(X_test) =)

The correct solution obviously is binary classification. For the comment above that says that binary classfication doesn't returns a probablity (for example SVM(classification) only returns a class and logistic, RFClassifier, XGBoostClassifier gives a probability and also a class given a threshold), you should ask yourself if that a regressor model returns always a probability, that is, if there is a restriction in a regressor model to predict values only in [0,1].

The Specialist is trying to determine whether a given transaction is fraudulent or not, which is a binary outcome (yes or no). Therefore, the problem should be framed as binary classification. The goal is to predict the probability of a transaction being fraudulent or not, and based on that, the Specialist can make a binary decision (fraudulent or not).

This is just binary classification, I don't understand how it could be anything else

It's B. This business problem can be framed as a binary classification problem, where the goal is to predict whether a given transaction is fraudulent (positive class) or not fraudulent (negative class). The model should output a probability for each transaction, indicating the likelihood that it is fraudulent.

should be D

Logistic regression models the probability of the default class (e.g. the first class). For example, if we are modeling people’s sex as male or female from their height, then the first class could be male and the logistic regression model could be written as the probability of male given a person’s height.

I think the answer is B, fraud has various cases which hard to define. So, Classification result will be fraud or not fraud. If Multi-category classfication, must define case of fraud in detaily

I believe the answer is B: it a binary classification problem because we are classifying an observation into one of two categories and the target variable in this problem is limited to two options: fraudulent or not fraudulent

well, regression classification is bullshit, I hope they formulate their questions better on the real exam. binary classification gives probability between 0 and 1