ENH NonBLASDotWarning -> EfficiencyWarning; Improve error message
DOC Add exceptions module to modules/classes.rst
MAINT Move ConvergenceWarning, UndefinedMetricWarning et al into exceptions
MAINT Remove ChangedBehaviorWarning from base
DOC/FIX Improve DataConversionWarning's docstring
Passing 1D arrays to check_array, without setting `ensure_2d` to false now
raises a deprecation warning before reshaping it. This will later throw an
error.
All Scaler classes also throw warnings when 1D arrays are passed.
All unit tests/doctests are modified to ensure that no 1D arrays are passed,
except in explicit 1D array tests where the warnings have been silenced.
Additional tests are also included which check for different 1D array cases.
2D array tests with one samples and one features are also added and where
they failed, `check_array` call has been modified to give a more useful error
message
Adding test for decision_function on sparse matrices
Fixing typos
Fix Memory Leak
PEP8 fixes
COSMIT PEP8
Fixed class_weight_label problem
Fixed label problem in tests
renamed label_ to classes_
Fixed OneClass bug
Adding kernel tests for sparse matrices
Pass dense data to decision function
Don't check precomputed kernel
use safe_sparse_dot in test, regenerate cython
create a core-dump by testing the binary case... hurray?
This commit adds support for a new libsvm parameter, which permits setting a
hard limit on the while loop in the svm.cpp file, in the Solver::Solve
function. Without this functionality, it has been observed that libsvm hangs
without converging in a reasonable time -- it can take days instead of
seconds, and perhaps it is in fact an infinite loop, I don't know. This
commit allows the user to set a hard limit on the number of libsvm iterations,
and modifies the libsvm bindings to raise a new exception of SolverTimeout
when this limit is reached.
Liblinear bindings removed from LinearSVC and LogisticRegression
prediction code in favor of NumPy code.
* no more copying of X at predict time
* no more unchecked malloc at predict time
* -4000 lines of C code
* simpler handling of coef_ and intercept_
Also, support symbolic class names and classes_ on both estimators.
seales