As the Lickert scale is categorical data, various ML approaches are possible. You might take a look RapidMiner Auto Model. The Likert scale is a multiclass classification which is possible. You could also turn your dataset into two classes e.g. 1, 2 vs the rest depending on whether you are using a 3-5 or 7 categorical scale.

As the Lickert scale is categorical data, various ML approaches are possible. You might take a look RapidMiner Auto Model. The Likert scale is a multiclass classification which is possible. You could also turn your dataset into two classes e.g. 1, 2 vs the rest depending on whether you are using a 3-5 or 7 categorical scale.

Yes, it can be processed via ML-Algorithms e.g., via Regression models.

There are two schools of thought. One is that Likert scale data is ordinal and therefore must be modelled using methods that only return values 1,2,3,4 or 5. The other is that you can assume that Likert data is continuous in the range 1 to 5 (or 1 to 7) and use standard parametric regression techniques.

There is a good article here that provides detailed justification for the second viewpoint:

Article LIkert scales, levels of measurement adn the “laws” of statistics

This article reports on detailed experiments using different machine learning techniques on Likert scale data, with evidence that Random Forests give good results:

Article Five statistical models for Likert-type experimental data on...

I would advise you to run your own experiments on your own data (where you know the actual values being predicted and hence can compare methods) and then form your own opinion.