Pros/Cons of Machine Learning Algorithms

I wouldn't pay attention to these groupings. They are artificially grouping on arbitrary factors and paying attention to the pros/cons could lead you astray.

For example: Regression shouldn't be it's own group because regression is commonly used in many decision tree algorithms. This group would be more apt to be described as 'linear models'. There are many different types of dimensionality reduction algorithms that are very different from LDA or PCA. The ensemble group are mostly decision tree algorithms and while technically they are ensembles, that is not what is thought of by most DS. AND, the NN categories show some of the elements of an algorithm (back-prop for example) as being an actual algorithm. I could go on...

"Regression" shouldn't be a separate category of algorithms given that the other categories are learning algorithms like ANNs, SVMs etc which are also used for regression. Also, it mentions Bayesian algorithms are fast - that's not true except for some specific ones.