khesui / fpmc Goto Github PK

If I have baskets with multiple items, what is the best way to represent this for input? For example, Figure 1 from the FPMC paper, I'd like to input User 1 (below).

Would that input look like:

1 a b a
1 a b b
1 a c a
1 a c b
1 b b a
1 b b b
...etc

In other words, every combination of the transition items in the baskets.

Hello author, When I ran this code, why the ACC is always equal to zero?

How to perform prediction for each user?

Hello,

Thank you for putting together this implementation from the paper. I had a couple of questions if you'd be open to answering them.

1. Data Structure --> for the 'idxseq.txt' file, could you please better describe what this represents? The last product_index is the last "item" in that basket? Is each row a separate order for a specific customer in sequential order?

2. Prediction Output --> I noticed the run.py runs through and outputs an "evaluation". Is there a method to get prediction outputs per customer?

Thank you

Hello,

I'd like to use your FPMC implementation in a commercial project. Is there any license, which determines copyrights for this repository? If not (at least I didn't find it in repo), could you please add it?

Thanks

How might one go about changing this to predict next baskets of multiple items?

Hi Chi-Ruei Li,

thanks for the amazing implementation of the paper!

In the function load_data_from_dir in utils.py you use:
b_tm1 = list(set(l[1:-1]))
using a set you change the order of the sequence, additionally you remove repetition of the items.

I would instead use
b_tm1 = np.asarray(l)[1:-1].tolist()

What are your thoughts on this?

Thanks,

Giuseppina