Converting .csv file to .xes

Amar

Hello,
I downloaded some data from an ERP system and converted to .csv file. I am using 'Convert CSV to XES' plugin to convert .csv to .xes. However, I am getting the error as attached. Preview is also attached.
Can you help me to identify issues? There are 2 columns that has data type as 'discrete'. Is this a problem?

Regards,
Amar

Amar

As mentioned in one of other forums, I started Prom by typing in 'ProM651.bat' from command prompt. Now I am getting below error for all the rows. 

Regards,
Amar

Skipping event, could not convert [348467, I, KEY, Start, 4151504211, 8.00E+12, EKKO, ME21, 1, 24/02/2003, RENDES, 15:55:44]

Error: java.text.ParseException: Could not parse 24/02/2003

fmannhardt

Hi Amar,
do you still get the output NullPointerException in the ProM activity view? 
I remember that there was a similar bug report a while ago, but I could not reproduce it.

Would you be able to send me (f.mannhardt@tue.nl) the event log? Of course, I will treat it as confidential data and delete it once I figured out what is going wrong. I assume the issue might be with the timestamps or the discrete columns, so you could only include those.

Another option would be to try out the current nightly build:
http://www.promtools.org/doku.php?id=nightly
And see if the error has been fixed in the meantime.

Amar

Hello, I found couple of issues. If I do not import time/date, it works fine. Another thing to note is that name of the first column should not have a number.

Regards,
Amar

Amar

Btw, i have already sent mail to your id.

Regards,
Amar

fmannhardt

Hi Amar,
I received the file and I could find the bug causing the error. Thanks for your effort.

I just fixed it and in a few hours there should be a new version of the Log package available for the nightly builds version.

Unfortunately, I'm unable to fix it in the released version due to the ProM release policy.

" I found couple of issues. [..] Another thing to note is that name of the first column should not have a number."

I would be glad to hear of all the issues. I'm keen to fix them before the next ProM release. I don't understand what you mean with "name of the first column should not have a number".

Regards,
Felix

Amar

Hi Felix,
Thanks for the message. I am able to locate nightly version.
How do I install and use nightly version?

Regarding the other issue, if the first column name of the .csv file contains a numeric character, the conversion fails. For e.g., in the .csv file that I shared, has first column header as 'CHANGENR2'. I removed '2' from the header name and then it worked fine.

Regards,
Amar

kkoopen

Hi,

I'm very interested in this problem, because I'm dealing with a very similar problem. When trying to convert my csv datafile into a xes log file I do get the same kind of error:

Skipping trace 450011751400080, could not convert[880211731465080,

(EKKO-EKPO) Purchase Order created, KENE, 1,494, 05.08.2015 00:00:00

1,235.34, null, null, ST, 2200, 910B 50 5000 Parchet, 10650578, 12010874, USD, 1.07250, 1.81, null, null, null, null, null, null, null]

Error: java.text.ParseException: Could not parse 05.08.2015  00:00:00

As well as posted here, when not importing the date and time the import works fine. I used the following date Format() : dd.MM.yyyy HH:mm:ss.

I hope somebody could help me out here, it would be very much appreciated!

Regards,

K

JBuijs

Hi K,

Did you investigate this line in the input data? Does the data format differ? Or are there other possible issues in this line in the data?