Home » Introductory Guide – Factorization Machines & their application on huge datasets (with codes in Python)

# Introductory Guide – Factorization Machines & their application on huge datasets (with codes in Python)

• George Stefanopoulos says:

Nice and informative article!

Here is a paper on the theoretical foundations on Factorization Machines by Steffen Rendel from Osaka University for further reading, :
https://www.csie.ntu.edu.tw/~b97053/paper/Rendle2010FM.pdf

• Ankit Choudhary says:

Thanks George!

• David Olmo says:

this was superb!! thank for the explanation, helped me a lot 🙂

• Ankit Choudhary says:

Thanks David!

• Dr Pavan Kumar says:

Good article to learn about PLS (Principles of Least squares) with the help of Matrices.

As a pure statistician, I am not finding any difference from OLSR technique to this methodology except providing a way to have clear understanding by novances.

• Ankit Choudhary says:

Thanks Pavan!

• Sayak Paul says:

The article is really informative. I came to know about FMs and FFMs with this one. I wanted know what are author’s motivation behind incorporating FMs and FFMs. More specifically, what are some other relevant techniques that can be equally incorporated?

• Ashish Tripathy says:

Thanks for the complete explanation. I remember struggling with it during the AV click prediction competition. In the same lines, If i have around 300 variables in a sparse dataset of products each variable showing a product’s profit earned for each user. Can i use first eg. to actually carry out a clustering over the data set?

• Ankit Choudhary says:

Hi Ashish, Can you elaborate? I don’t understand what you want to do?

• jatinpal singh says:

Nice explanation at the start and informed about xlearn

• Ankit Choudhary says:

Thanks Jatinpal!

• Anshul says:

Nice read and good knowledge. According to statistics it is machine learning market will be close to \$23 BN by 2020

• Mayank says:

Should all the variables be categorical or every variable one-hot encoded?
For e.g.: A variable, say id, should it be one hot encoded or just converted to type “category” and then feeded to the model?