Add SVC documentation for properties. #5224
Conversation
Issue #4687. Documented `fit_status_`, `probA_`, and `probB_`.
| 0 if correctly fitted, 1 otherwise (will raise warning) | ||
|
|
||
| probA_, probB_ : array | ||
| probability estimates, empty array for probability=False |
There was a problem hiding this comment.
What are the shapes, and what are the interpretations?
|
thanks :) |
| parameters used to produce probability estimates from decision values, | ||
| empty array for probability=False. See the section about scores and | ||
| probabilities in the SVM section of the User Guide for details. | ||
|
|
There was a problem hiding this comment.
Also thought Section 8 of http://www.csie.ntu.edu.tw/~cjlin/papers/libsvm.pdf was useful. Should I include in docstring?
There was a problem hiding this comment.
Can you add it as References to the docstring? Does it also describe the SVR? Can you check that it is also mentioned in the user guide? Thanks.
There was a problem hiding this comment.
It's in the user guide: http://scikit-learn.org/stable/modules/svm.html#implementation-details.
There was a problem hiding this comment.
probA_ and probB_ are not mentioned in the user-guide, right?
There was a problem hiding this comment.
http://scikit-learn.org/stable/modules/svm.html#scores-probabilities explains Platt scaling, but doesn't explicitly mention probA_ or probB_.
|
Will render as shown below. Let me know if I should link to the LIBSVM paper differently.
|
|
It would be good to give a short explanation of what they do so the user doesn't have to look it up. They are the scaling and offset learned in the platt scaling, right? It would be good to give a formula of what they mean. Otherwise they are useless. |
|
@amueller, anything I should change on the last commit? |
| fit_status_ : int | ||
| 0 if correctly fitted, 1 otherwise (will raise warning) | ||
|
|
||
| probA_, probB_ : array, shape = [n_class * (n_class-1) / 2] |
There was a problem hiding this comment.
Each should be on its own line. Though you only need the explanation once.
Otherwise it looks greatk.
|
Sorry for the long turnaround |
|
finished in #12427 |
Issue #4687.
Documented
fit_status_,probA_, andprobB_.