I am very new to machine learning, and it's hard for me to know in what direction I should go. Thanks in advance for your help.

I have a set of data showing what customers bought:

  • A customer may have bought 1 or more items
  • Each item is discribed by :
    • A quality level which is an integer between 1 and 7
    • Its content from a first point of view (the item has x1% of content 1, x2% of content 2, ...)
    • Its content from a second point of view (the item has xA% of content A, xB% of content B, ...)
  • A customer says that he belongs to the group G1, G2 or G3

I would like to be able to guess the group a new customer belongs to, given the items he bought.

What algorithm should I ideally use ? What algorithm would you recommend if I can only use numpy, without scipy or scikit (it's on a server which only has numpy) ?

Thank you very much

Edit : I'm adding the appropriate words to the title and to the tags


1 Answer 1


As per the problem statement statement it does look like 3 class classification problem. But I don't think you can use any ML algorithms using Python without sklearn package in it. Try getting it installed and you can implement the problem using Logistic regression, Decision Tree, there are still more classification algorithms. However if you are planning to extract a pattern from the dataset and making into cluster, you can use K-means clustering which is a un-supervised algorithm.

  • $\begingroup$ Thanks to your comment, I now have some useful keywords. I found a website with different algorithms implemented from scratch, and in particular the Naive Bayes : machinelearningmastery.com/… However, I am not sure of the properties caracterizing the customers (i don't really have a table with a line for each customers, and a column for each property) : how to choose these properties ? $\endgroup$
    – Corlin
    May 6, 2018 at 17:52
  • $\begingroup$ @Corlin I hope by this time you might have already solved the problem. If not could you please check if it is possible to paste the head of the dataset you have ? $\endgroup$
    – deepguy
    Jan 9, 2019 at 16:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.