Cumulative net worth - different values between dataset in xes and csv

Lorenzo

I have been using both the .xes and .csv dataset, available on the challenge webpage, on different software.

The .xes, when opened in ProM, and filtered on event attributes, shows a maximum Netw worth value of 28.405.633,00 EUR Milion, that is 2,84*10exp7. It refers to Logistic services, Road Packed, (Purch. Doc. 4507000684, item 10, Remove Payment Block).

The .csv file, when opened with different commercial software (if the dataset is not a built-in by the software house) shows a much bigger figure.

I checked the table connecting the .xes file to PowerBI and it shows for the same trace a value of 2,84*10exp14. The table needs a transofrmation, because net worth figures are with a decimal zero that is converted into integer, so there is a zero to be taken away, but still the value is 2,84*10exp13.
By arranging the Net Worth column on descending filter, it shows that there are weird figures for the following purchase doc.: 4507000684, 4507000430, 4507001930,  4507004994.

We could filter out those traces, nonetheless, it is advisable to double-check the .csv and .xes.

Lorenzo

I must correct myself: I checked the .csv file with PowerBI (not the .xes). Findings remain anyway.

MrEater

Hi Lorenzo,

I just checked both files for the event maximum Netw worth value of Purch. Doc. 4507000684 - and the values are identical: 2.8405633E7.

Maybe double-check the transformations you mentioned.

Lorenzo

Thank you for your answer.  First, my fault: the maximum value is 2.89945303E7(EUR, Purch. Doc 4507004994). This is correct in ProM with *xes and other software (Celonis and Disco) with their prebuilt dataset) as well as other softwares (TimelinePI, Minit and Lana uploading the *csv file.  That is or the process mining.
Passing to some analytics, reading the *csv file on Alteryx, I get the same result. 
If I read the *csv file with PowerBI, I get those weird figures as mentioned in my first comment.
It is not a big issue, for the purpose of mining the process. It could be if using different ools to extend the range of visualization you want to achieve.