Are K-means clustering and Logistic Regression comparable?

[shibbir63]

Hi,

I will be grateful if you could kindly tell me  if the outcomes of K-means clustering and logistic regression are comparable, like we compare classification decision tree with logistic regression for model performance.

Regards,

Shibbir

[PaigeMiller]

"Comparable" is a relatively vague and subjective term, and even though you provide an analogy, the exact analogy really doesn't mean anything to me.

How would you propose we compare K-means and Logistic regression? On what criteria should they be compared?

[shibbir63]

Hi,

Apologies for the confusion. 

If both K-means and Logistic regression are applied to the same data set, are their performance comparable to determine which model to choose as final model, or they are used to model different kind of outcomes and as a result not comparable.

For example, we can compare Decision tree with Logistic regression to assess their relative performance based on confusion metrics, ROC curve, errors etc. to determine which model perfumed well and be chosen.

I am a new learner in this space, so a little insights will be great. So far I have not come across any example during the last few weeks related to comparing K-means and logistic regression.

Kind regards,

Shibbir

[PaigeMiller]

Totally agree with @Ksharp . These two methods cannot be compared.

[Ksharp]

No. They are totally different model.
K-means is NON-Supervise learning model (doesn't have Y),
while Logistic is Supervise learning model (have Y).

K-means can't be used to predicting model, but Logistic could .

[shibbir63]

Many thanks @Ksharp .

And what about K-means, and Association/Market Basket Analysis? Are they also completely different?

Kind regards,

Shibbir

[PaigeMiller]

---

@shibbir63 wrote:
Many thanks @Ksharp .

And what about K-means, and Association/Market Basket Analysis? Are they also completely different?

---

Ask yourself the question: what types of data do these work on? Does not that provide the answer?

by the way ... you really need to avoid subjective terms like "completely different" and actually state the criteria you want to use, because you really want to evaluate the procedures by certain criteria, whereas other criteria are probably not relevant. For example, both K-means and Association/Market Basket analysis are tools that work on data, so they are not completely different, but this is probably irrelevant.

[shibbir63]

Thanks PaigeMiller.

[Ksharp · Accepted solution]

Market Basket Analysis actually is not a model, just a summary statistic method.
But it is NON-Supervise learning model like K-means .

[shibbir63]

Many thanks Ksharp.

The clarification is really helpful in my learning .

Regards,

Shibbir