Ticket #1253 (closed bug: fixed)
Error in probability estimator constructor argument parsing
Reported by: | ales | Owned by: | janez |
---|---|---|---|
Milestone: | Component: | library | |
Severity: | minor | Keywords: | |
Cc: | matija | Blocking: | |
Blocked By: |
Description
ProbabilityEstimatorConstructor_call does not parse the argument tuple correctly when using None for the distribution (as documented here)
import Orange iris = Orange.data.Table('iris') Orange.statistics.estimate.M(None, iris) # raises TypeError: Invalid arguments for 'ProbabilityEstimatorConstructor.call' prior = Orange.statistics.distribution.Distribution("iris", iris) Orange.statistics.estimate.M(None, prior, iris) # raises TypeError: Invalid arguments for 'ProbabilityEstimatorConstructor.call'
Is this even worth fixing? Considering that no builtin constructors actually use the instances argument. Instead we could just simplify the documented interface (leave out the instances and weight_id parameters). Instead have helper functions in the estimate package that compute estimators from Orange.data.Table.
Attachments
Change History
comment:1 Changed 17 months ago by janez
Aleš, I agree: we shouldn't waste time with this. Please simplify the documentation or do whatever is the easiest.
comment:2 Changed 14 months ago by ales
- Status changed from new to closed
- Resolution set to fixed
Fixed in da96782f1c2b and 02feeae55f5f
Note: See
TracTickets for help on using
tickets.