學生對 Coursera Project Network 提供的 Hierarchical Clustering: Customer Segmentation 的評價和反饋

This is a beginners level course. The instructor uses scipy and sklearn libraries without explaining them.

The project scope could have been expanded like adding other hierarchical clustering methods and then maybe differentiating it and then justifying which would be a better choice. However, this project basically highlights the application of hierarchical clustering using a dendrogram.

This course is unrealistically simplified and the project contains around 20 lines of codes in total.

Hi there, I must say that this project does not really have the depth that I would have like to receive with respects to clustering. The host did not talk about scaling each of the features, the choice of distance metric and what effect that has, and furthermore, what ward actually means, aside from minimising variance. In addition, there was not any information on understanding what the clusters actually mean - the fundamental problem in unsupervised learning. Perhaps I am asking too much from a beginner course but I am left with code, but little understanding.

That is a added vlue the profil, a trick to have in unsupervised learning. You would like the short just right of the work but the explaination is lack about libraries so it is to look about ;)

Perfect Project. Thank You

This is very good course

amazing experience

Very Good

very easy

very good

Nice

good

Good course, but the instructor was hard to understand and the quiz did not test your knowledge well. One reccomendation I would have is to add a coding assignment so that I know whether I understand the material.

Excellent Project . But We Cant Understand Well by this videos ;)

Unfortunately, the functions was not fully explained to student. Maybe it can be supported with more visual materials or can be repeated with different examples to understand difference. Questions were left on minds such as why we chose those parameters and what other options could be, what is the way we want to go more detailed analysis.

Moreover audio quality was bad and interface that used for coding is not handy for low resolution pc's.

Little explanation of how/why clustering actually works

worst sound quality