Math error

Bug #1095293 reported by Paweł Bednarz
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
BNFinder
New
Undecided
Unassigned

Bug Description

There is something wrong when executing bnf (bnf -e training_set.txt -c test.cpd) on data as in the attachment. When the last value for dayl is changed (e.g. for 16.2) everything executes properly. Trace log:

  File "../bnf", line 66, in <module>
    fpr=options.fpr,)
  File "/home/pawel/Projects/bnfinder_help/trunk/BNfinder/data.py", line 677, in learn
    v_signs=self.select_1(v,par).signs()
  File "/home/pawel/Projects/bnfinder_help/trunk/BNfinder/data.py", line 160, in signs
    sign = stats.pearsonr(self.data[i],self.data[-1])[0]
  File "/home/pawel/Projects/bnfinder_help/trunk/BNfinder/stats.py", line 247, in __call__
    return apply(self._dispatch[type(arg1)], (arg1,) + args, kw)
  File "/home/pawel/Projects/bnfinder_help/trunk/BNfinder/stats.py", line 821, in lpearsonr
    t = r*math.sqrt(df/((1.0-r+TINY)*(1.0+r+TINY)))
ValueError: math domain error

Revision history for this message
Paweł Bednarz (pawel-bednarz) wrote :
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.