[python] convert datatable to numpy directly #1970

guolinke · 2019-01-26T03:13:30Z

fix #1968

StrikerRUS · 2019-01-26T04:24:00Z

What about slicing?
https://github.com/Microsoft/LightGBM/blob/b37065dbd5510079aac341a34895f4854c782fcb/python-package/lightgbm/basic.py#L1397-L1403

I might be wrong, but I suppose that sklearn utils cannot work with DataTable:
https://github.com/Microsoft/LightGBM/blob/b37065dbd5510079aac341a34895f4854c782fcb/python-package/lightgbm/sklearn.py#L482-L486
https://github.com/Microsoft/LightGBM/blob/b37065dbd5510079aac341a34895f4854c782fcb/python-package/lightgbm/sklearn.py#L598-L599

Is it "official" feature? If so, I can help with docstings updates, e.g.
https://github.com/Microsoft/LightGBM/blob/b37065dbd5510079aac341a34895f4854c782fcb/python-package/lightgbm/basic.py#L646

guolinke · 2019-01-26T07:21:30Z

Thanks @StrikerRUS , I push some fixes.

guolinke · 2019-01-26T07:22:16Z

python-package/lightgbm/basic.py

@@ -1398,6 +1402,8 @@ def get_data(self):
                self.data = self.data[self.used_indices, :]
            elif isinstance(self.data, DataFrame):
                self.data = self.data.iloc[self.used_indices].copy()
+            elif isinstance(self.data, DataTable):
+                self.data = self.data[self.used_indices, :]


I am not very sure about here, should we need a copy or not?

Me too... To be honest, I see DataTable for the first time 😄

Speaking about pandas, I used .copy() with the aim to prevent SettingWithCopyWarning in users' preprocess functions.

Maybe address this question to someone from h2o?

Okay, I see. So theoretically, we don't need a copy here.
@pseudotensor any idea about this?

ping @pseudotensor
Please take a look at this function for getting raw data.

python-package/lightgbm/sklearn.py

StrikerRUS · 2019-01-26T10:54:25Z

I've updated more docstrings and simplified isinstance check.

python-package/lightgbm/compat.py

guolinke · 2019-01-30T05:55:03Z

@StrikerRUS I think we can merge this, how do you think?

StrikerRUS · 2019-01-30T13:36:17Z

@guolinke I'm just worrying about the need of copy...

guolinke · 2019-02-03T14:08:50Z

@StrikerRUS I searched 'copy' in datatable documents, and didn't find anything.
https://datatable.readthedocs.io/en/latest/search.html?q=copy&check_keywords=yes&area=default
I think they don't need this.

We also can merge this first, and provide the fix for that in the future if needed.

StrikerRUS · 2019-02-03T16:34:31Z

@guolinke Agree!

Seems that numpy and DataTable slicing cases can be merged, if we don't need to copy:

if isinstance(self.data, (np.ndarray, DataTable)) or scipy.sparse.issparse(self.data):
    self.data = self.data[self.used_indices, :]

guolinke · 2019-02-04T07:01:59Z

I prefer to separate them, for the potential future needs.

#2) * [ci] removed temp brew hotfix and deprecated sudo option (microsoft#1951) * removed brew hotfix and deprecated sudo option on Travis * removed brew hotfix on Azure * updated Boost docs (microsoft#1955) * removed warnings about types in comparison ([-Wsign-compare]) (microsoft#1953) * removed comparison warning * fixed spacing * [docs] ask to provide LightGBM version for issue (microsoft#1958) * [R] Fix multiclass demo (microsoft#1940) * Fix multiclass custom objective demo * Use option not to boost from average instead of setting init score explicitly * Reference microsoft#1846 when turning off boost_from_average * Add trailing whitespace * [R] Correcting lgb.prepare output comment (microsoft#1831) * Correcting lgb.prepare output comment * updated Roxygen files * [docs] bump xcode version in docs (microsoft#1952) * fix typo * [docs] Added the links to the libraries used (microsoft#1962) * Added links to the libraries used. * Fixing the header * Fixes * ot -> to * [docs] fixed minor typos in documentation (microsoft#1959) * fixed minor typos in documentation * fixed typo in gpu_tree_learner.cpp * Update .gitignore * support to override some parameters in Dataset (microsoft#1876) * add warnings for override parameters of Dataset * fix pep8 * add feature_penalty * refactor * add R's code * Update basic.py * Update basic.py * fix parameter bug * Update lgb.Dataset.R * fix a bug * Fix build on macOS Mojave (microsoft#1923) * Fix build on macOS Mojave Fixed microsoft#1898 - https://iscinumpy.gitlab.io/post/omp-on-high-sierra/ - https://cliutils.gitlab.io/modern-cmake/chapters/packages/OpenMP.html - Homebrew/homebrew-core#20589 * update setup.py * update docs * fix setup.py * update docs * update docs * update setup.py * update docs * [tests][python] added tests for metrics' behavior and fixed case for multiclass task with custom objective (microsoft#1954) * added metrics test for standard interface * simplified code * less trees * less trees * use dummy custom objective and metric * added tests for multiclass metrics aliases * fixed bug in case of custom obj and num_class > 1 * added metric test for sklearn wrapper * [python][R][docs] added possibility to install with Visual Studio 2019 Preview (microsoft#1956) * Found error from microsoft#1939 (microsoft#1974) * fix more edge cases in mape (microsoft#1977) * fix R's overflow (microsoft#1960) * [tests][python] added test for huge string model (microsoft#1964) * added test for huge string model * fixed tree sizes field * simplified model structure * fixed test and added try/except * fix nan in eval results (microsoft#1973) * always save the score of the first round in early stopping fix microsoft#1971 * avoid using std::log on non-positive numbers * remove unnecessary changes * add tests * Update test_sklearn.py * enhanced tests * fix microsoft#1981 * [python] added OpenMP options for python-package installation (microsoft#1975) * added OpenMP options for python-package installation * fixed grammar typo * improved model loading routines (microsoft#1979) * [ci] refined command status check (microsoft#1980) * refined command status check * refined Appveyor * redirect all warnings to stdout * cpplint whitespaces and new lines (microsoft#1986) * fix microsoft#1994 * [docs] Fixed OpenCL Debian package name typo (microsoft#1995) [docs] Fixed OpenCL Debian package name typo * [python] convert datatable to numpy directly (microsoft#1970) * convert datatable to numpy directly * fix according to comments * updated more docstrings * simplified isinstance check * Update compat.py * [R-package] Fix demos not using lgb.Dataset.create.valid (microsoft#1993) * Hand edit broken commit * Hand edit broken commit * Hand edit broken commit * Hand edit broken commit * 2.2.3 release (microsoft#1987) * Update DESCRIPTION * Update DESCRIPTION * update version number at master branch (microsoft#1996) * Update VERSION.txt * Update .appveyor.yml * Update DESCRIPTION * Initial attempt to implement appending features in-memory to another data set The intent is for this to enable munging files together easily, without needing to round-trip via numpy or write multiple copies to disk. In turn, that enables working more efficiently with data sets that were written separately. * Implement Dataset.dump_text, and fix small bug in appending of group bin boundaries. Dumping to text enables us to compare results, without having to worry about issues like features being reordered. * Add basic tests for validation logic for add_features_from. * Remove various internal mapping items from dataset text dumps These are too sensitive to the exact feature order chosen, which is not visible to the user. Including them in tests appears unnecessary, as the data dumping code should provide enough coverage. * Add test that add_features_from results in identical data sets according to dump_text. * Add test that booster behaviour after using add_features_from matches that of training on the full data This checks: - That training after add_features_from works at all - That add_features_from does not cause training to misbehave * Expose feature_penalty and monotone_types/constraints via get_field These getters allow us to check that add_features_from does the right thing with these vectors. * Add tests that add_features correctly handles feature_penalty and monotone_constraints. * Ensure add_features_from properly frees the added dataset and add unit test for this Since add_features_from moves the feature group pointers from the added dataset to the dataset being added to, the added dataset is invalid after the call. We must ensure we do not try and access this handle. * Remove some obsolete TODOs * Tidy up DumpTextFile by using a single iterator for each feature This iterators were also passed around as raw pointers without being freed, which is now fixed. * Factor out offsetting logic in AddFeaturesFrom * Remove obsolete TODO * Remove another TODO This one is debatable, test code can be a bit messy and duplicate-heavy, factoring it out tends to end badly. Leaving this for now, will revisit if adding more tests later on becomes a mess. * Add documentation for newly-added methods. * Initial work towards add_data_from This currently only merges the feature groups and updates num_data_. It does not deal with Metadata or non-dense bins yet. * Fix bug where dense bin copy of num_data_ wasn't updated * Small bug fix in dense_bin.hpp, initial implementation of Merge for 4-bits bin. * Add unit test for dense bin case of add_data_from, and refactor tests slightly. * Initial implementation of Merge for sparse bins and unit tests for it. * Ensure we test merging sparse data sets after loading them from binary This seems silly, but push_buffers_ aren't populated if the data was loaded from a binary file. This forces us to reconstruct the index,value form of the data in the target bin before merging. Adding this test ensures that code is covered. * Add labels to text dumps. * Add weights to text dumps. * Ensure add_data_from properly merges labels. * Ensure metadata appends weights correctly, and unit test for it. * Implement metadata merging for query bits This is currently not covered by unit tests. * Check datasets are aligned before merging. This catches the majority of obvious errors, e.g. not having the same number of features or having different bin mappings. * Add test that booster behaviour is preserved by add_data_from. * Add configuration parameters for CEGB. * Add skeleton CEGB tree learner Like the original CEGB version, this inherits from SerialTreeLearner. Currently, it changes nothing from the original. * Track features used in CEGB tree learner. * Pull CEGB tradeoff and coupled feature penalty from config. * Implement finding best splits for CEGB This is heavily based on the serial version, but just adds using the coupled penalties. * Set proper defaults for cegb parameters. * Ensure sanity checks don't switch off CEGB. * Implement per-data-point feature penalties in CEGB. * Implement split penalty and remove unused parameters. * Merge changes from CEGB tree learner into serial tree learner * Represent features_used_in_data by a bitset, to reduce the memory overhead of CEGB, and add sanity checks for the lengths of the penalty vectors.

convert datatable to numpy directly

0397edf

fix according to comments

056fdf1

guolinke commented Jan 26, 2019

View reviewed changes

StrikerRUS reviewed Jan 26, 2019

View reviewed changes

python-package/lightgbm/sklearn.py Outdated Show resolved Hide resolved

StrikerRUS added 2 commits January 26, 2019 13:48

updated more docstrings

c8317db

simplified isinstance check

ee369c8

StrikerRUS changed the title ~~convert datatable to numpy directly~~ [python] convert datatable to numpy directly Jan 28, 2019

StrikerRUS reviewed Jan 28, 2019

View reviewed changes

python-package/lightgbm/compat.py Outdated Show resolved Hide resolved

Update compat.py

9e4f6d2

StrikerRUS mentioned this pull request Feb 2, 2019

2.2.3 release #1987

Merged

StrikerRUS approved these changes Feb 4, 2019

View reviewed changes

guolinke merged commit 2c9d332 into master Feb 4, 2019

guolinke deleted the data-table-support branch February 5, 2019 00:35

lock bot locked as resolved and limited conversation to collaborators Mar 11, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[python] convert datatable to numpy directly #1970

[python] convert datatable to numpy directly #1970

guolinke commented Jan 26, 2019

StrikerRUS commented Jan 26, 2019

guolinke commented Jan 26, 2019

guolinke Jan 26, 2019

StrikerRUS Jan 26, 2019 •

edited

Loading

guolinke Jan 27, 2019

StrikerRUS Jan 30, 2019

StrikerRUS commented Jan 26, 2019

guolinke commented Jan 30, 2019

StrikerRUS commented Jan 30, 2019

guolinke commented Feb 3, 2019

StrikerRUS commented Feb 3, 2019

guolinke commented Feb 4, 2019

[python] convert datatable to numpy directly #1970

[python] convert datatable to numpy directly #1970

Conversation

guolinke commented Jan 26, 2019

StrikerRUS commented Jan 26, 2019

guolinke commented Jan 26, 2019

guolinke Jan 26, 2019

Choose a reason for hiding this comment

StrikerRUS Jan 26, 2019 • edited Loading

Choose a reason for hiding this comment

guolinke Jan 27, 2019

Choose a reason for hiding this comment

StrikerRUS Jan 30, 2019

Choose a reason for hiding this comment

StrikerRUS commented Jan 26, 2019

guolinke commented Jan 30, 2019

StrikerRUS commented Jan 30, 2019

guolinke commented Feb 3, 2019

StrikerRUS commented Feb 3, 2019

guolinke commented Feb 4, 2019

StrikerRUS Jan 26, 2019 •

edited

Loading