aliannejadi / lsapp Goto Github PK

I have calculated statistics for app usage records, and it does not match with statistics from readme (and paper). Other statistics (sessions, unique apps, users, mean duration/user) matches with numbers from table in readme file.

Also I have observed that (train, validation, test) instances from statistics table does not sum up to app usage records statistic (663,954 vs 599,635).

I have tried to cleanse data with several ways, but could not obtain a number even close to one I mentioned above.

Records filtering method mentioned in paper:

we considered all repeated app usage records with less than one minute time difference as one
record.

I have tried to merge only consecutive app records as in quote and got 692,217 apps usage record. When I merged with ignoring consecutive requirement, I got 346,549 records.

Questions:

1. Why do not (train, validation, test) instances sum up to number of app usage records?
2. What are the cleansing methods should be applied to obtain same numbers as you put in data statistics table?
3. Some users have small number of records (less than 10), they can't be splitted to (train, validation, test) instances. Did you filter (remove) those users?

Hi, can I use your dataset for my company's project? I couldn't find any information about the license for your dataset.
Thank you