Exam AWS Certified Machine Learning - Specialty topic 1 question 89 discussion

A is correct. tSNE can do segmentation or grouping as well. Refer: https://towardsdatascience.com/an-introduction-to-t-sne-with-python-example-5a3a293108d1

A is definitely the correct answer. Pay attention to what the question is asking: "whether there are natural groupings for these columns across all customers and visualize the results as quickly as possible" The key point is to visualize the "groupings"(exactly what t-SNE scatter plot does, it visualize high-dimensional data points on 2D space). The question does not ask to visualize how many groups you would classify (K-Means Elbow Plot does not visualize the groupings, it is used to determine the optimal # of groups=K).

option A

B doesn't even answer the question: how are you going to see your customer groups in an elbow plot

Euclidean Distance suffers for high dimensional data. tSNE can suffers as well, but from my perspective is the correct one.

Elbow plot will not help visualize groups, only try to predict an optimal number of clusters. I think A is a better choice here

A. The t-SNE algorithm is a popular tool for visualizing high-dimensional datasets, as it can transform high-dimensional data into a 2D scatter plot, which makes it easier to visualize and understand the relationships between data points. The scatter plot produced by t-SNE can be interpreted as a map that reveals the structure of the data, showing whether there are natural groupings or clusters within the data. Option A is the quickest and simplest way to visualize the data in a meaningful way, allowing the Specialist to gain insights into the data more efficiently.

A is correct

12-sep exam

A as k-means elbow is erroneous. It does not helping here. Scatter plot and t-sne is the right answer

An elbow plot (B) will not give you what the question is asking for. A scatter plot will, and t-SNE is first for visualizing before dimensionality reduction.

A is correct as k means suffer from curse of dimensionality and t-she will be a better option.

The B,C,D plots are meaningless wrt the problem —> A

t-SNE suffers curse of dimensionality and is indicated for small datasets

Additionally the numeric features don't require "embedding". I think they meant to write "standardize"

Rooting for A

B & D are wrong--because data contains "thousands of columns" and using k-means with euclidean suffers from "curse of dimensionality" Thus leaving A & C, you CANNOT viz clusters/groups/segments in a line graph so correct answer is A