# How to choose the best algorithm

this might be simple but I need to know how to choose a best algorithm based on a scenario.

I have a dataset. The target class is, let's say color, this color attribute can have upto 5 values (red, yellow, blue, green, pink)

So I have:

1.) 1 target class - color

2.) Multiple values for that target class - (red, yellow, blue, green, pink)

As this is a labelled dataset with known classes, I can use a Supervised Learning Algorithm

But I don't know which algorithm would be the best. It should be a multi-class classigication? Or multi-labeled ? Can someone help me to find the best algorithm?

note: I know choosing an algorithm depends on many factors (such as its acuracy), but at a glance, what type of algorithms should I try?

Thank you

Okay, so without going in to details,

Your target values are nominal, discrete values, hence the classification model would solve your problem.

Under classification there are variety of methods available,for example, there is decision tree, Naive Bayes etc.. . you may try all of these and check which gives the better confusion metrics.

Multi-Class means, there are multiple classes , such as in you colors.

If I have books features as input and i want to predict its genre, and the available categories i can put this book in are- fictional, romance, historical. all these labels makes it multi class because multiple genre.A book can be of any of the genre.

Multi-Label means, an instance can belong to more than 1 class.

Now, if this one book is fictional romance , then this makes it multi label. Because out of 3 available categories, this one book belongs to 2(it can be more than 2 also).

Now, This was just a short answer to get you started,please read on these key words to further details.

1.) 1 target class - color

This is target attribute, color

2.) Multiple values for that target class - (red, yellow, blue, green, pink)

Multiple class/labels for target attribute, red, yellow, blue... are classes/labels. Since there are more than 2 classes its multi class.

If you input map to 2 or more colors than 1 then it makes it multi label.

• Thank you for the explanation. So multi-class means an attribute having multiple values and multi-label means the target can be multiple attributes. Am I right? May 22 '20 at 16:30
• just added more clarification with example in answer. May 22 '20 at 17:17
• Thank you now it make even more sense. May 23 '20 at 6:33
• Happy to help. Accepting the answer would be nice. May 23 '20 at 8:39