Request for simple example data and script.

[swayson]

Hi there,

Recently read your paper and really great work. I wanted to give the code a try. At the moment, I tried to follow the example in the manual you created but seem to get NoneType errors. Haven't spend time digging into the code, to exactly what is going on, but wanted to try my luck and ask you if you could perhaps create a simple input file and example script to toy around with?

I tried this:

The input file:

0,1,A,B
0,2,A,C
1,1,B,D
1,4,B,C

This gave no errors.

$ python events.py /tmp/data.csv /tmp/input.sqlite -t 0 -w 1

But here is where things went wrong.

python dynsnap.py /tmp/input.sqlite input.out -p -w -1
# file loaded: /tmp/input.sqlite
# Total time range: None None
# Columns: tlow thigh dt sim number_of_events
Traceback (most recent call last):
  File "dynsnap.py", line 1260, in <module>
    main(argv=sys.argv[1:], return_output=False)
  File "dynsnap.py", line 1160, in main
    x = finder.find()
  File "dynsnap.py", line 480, in find
    if self.tstart >= self.tstop:
TypeError: unorderable types: NoneType() >= NoneType()

[rkdarst]

Hi,

I've finally looked into this and here's my analysis:

First, the code uses space separations, not comma separations. So when it tried to run, it did nothing but create an empty DB. Then, events.py acts as a cache system: if it sees there is already a DB there, it does nothing. So future runs probably returned silently without giving note that the DB was empty.

Then, when you run dynsnap.py, it fails because there are no events. In this case, it happened to use "None" as the start and end times. In python2, these are orderable, but you get the error you saw in python3. I've tried to make things python3 compatible, but I guess that path wasn't tested.

I've done the following:

• Added a --separator option, which can be used to specify ',' as a separator. Note that it's a dumb .split().
• Added a check, dynsnap.py will raise an error if there are no events within the database. This should provide error checking so it's obvious if an import failed.

Is this a proper analysis?

I've made some new commits now but haven't fully tested, I should in the next few days.

Thanks for the good report,

• Richard

[LanlanYu11]

hello,
Recently I read your paper and tried to follow the example in the your paper. I got the same result, when I used the data of the Enron_mail. But unfortunately, use the Reality mining, the result is different.
I have data from the personal devices of the 94 subject, contains 759914 unique readings from 2004/1/1 until 2005/7/14. Use the day as unit , inconceivably, I got the four time slices.....
`python dynsnap.py input.sqlite input.out -w -1

file loaded: input.sqlite

Total time range: 0 560

Columns: tlow thigh dt sim number_of_events

0 330.0 330.0 0.321510498169 5010
330.0 430.0 100.0 0.345930733212 3489
430.0 540.0 110.0 0.312115168681 1382
540.0 560.0 20.0 0.00175915835 14`
I really need your help.
Thanks.