hzhaoaf
commented
6 years ago @SeverusBing Which part are you running when you encounter this error?
In fact, the parameter "data" in the init functions is the union set of training data and test data. You may check Line 234 in method "run_baseline", which show you how to give the input parameters.
For train_data and test_data, their format are triplets, i.e., (user_id, item_id, rating). You can check the data in "_data/yelp-50k/exp_split/1/ratings_train1.txt" when you unzip the _FMG_releaseddata.zip in this project.
The case when "data=None" is an early-stage code for me to test other datasets. You can just ignore it actually.
SeverusBing
Hi Dr. Zhao, I performed the latest code recently and I finally stopped by an input error in the file 'mf.py'.
There is a class 'MF_BGD' in 'mf.py' and the attribute 'init' of it needs a parameter 'data', when this input parameter is 'None', the function will read a data file called 'data/ml-1m-rating.txt'. I found that there was no such file in your data package so I wonder if you mind showing an format example of the file here, cause I don't know which form to transform my data in or what kind of data I should use.
Also, I will be very appreciated if you can tell the details about all parameters of 'MF_BGD', such as 'train_data' and 'test_data'.
Thank you for your kindly help! I am looking forward to your reply.