Exam AWS Certified Machine Learning - Specialty topic 1 question 51 discussion

HOW MANY/MUCH, THOSE ARE REGRESSION TOPIC, LOGISTIC FOR 0/1,YES/NO https://docs.aws.amazon.com/zh_tw/machine-learning/latest/dg/regression-model-insights.html THE ANSWER SHOULD BE D.

Amazon SageMaker Random Cut Forest (RCF) is an unsupervised algorithm for detecting anomalous data points within a data set Answer is D 100%

The problem involves predicting the number of units to be produced each quarter based on historical sales data. This is a continuous numerical prediction, making it a regression problem. Linear regression is ideal for forecasting when there is a linear relationship between input variables (e.g., past sales, seasonal trends) and the target variable (units to be produced). It helps model the relationship between past sales and future demand. If there are seasonal effects, a time-series model (like ARIMA or Prophet) could be considered as well.

D should be the answer

How many units should give this away as Linear regression

I do not see any hint of anomalies here, we are looking for a number to be predicted, this seems to be the reason of the correct answer https://docs.aws.amazon.com/quicksight/latest/user/how-does-rcf-generate-forecasts.html

How can the right answer be B? That Random Cut Forest is an algorithm written for anomaly detection.

option D

D is the correct. B is for outlier detection only.

It sounds like Linear regression problem and Random Cut is more known for anomaly detection while it can do other types of ML. The answer seems to be strange with no explanation.

D is correct!

D. Linear regression would be the appropriate machine learning approach to solve this problem of predicting the number of units of a particular part to be produced each quarter. Linear regression is a supervised learning algorithm used for predicting continuous variables based on input features. In this case, the historical sales data can be used as input features, and the number of units produced each quarter can be used as the continuous target variable.

definitely D.

This is a regression problem where the goal is to predict a continuous outcome, which in this case is the number of units of a particular part that should be produced each quarter. Linear regression is a simple and commonly used approach to solve such problems, where a linear relationship is established between the independent variables (e.g., historical sales data) and the dependent variable (e.g., number of units of a part to be produced).

D, RCF answers here just link one article where RCF is implemented to find outliers in time series, or are able to deduce trends, but here they mention already labelled data, RCF is unsupervised, so that data would go to waste.

Honestly, i think these are all bad answers. It should be time series modeling methods.

The answer is D. B is out of it as RCF is used for anomaly detection. Logistic Regression is for Classification mainly. Only linear regression can be used if Time series algorithms are not part of the options.