[ISSUE-10463] Add missing import in learning-to-rank tutorial

[jpizagno]

Introduction

As mentioned in ISSUE-10463, the query id needs to be sorted. Following the Learning to Rank tutorial, one gets the following error: Check failed: non_dec: qid must be sorted in non-decreasing order along with data.. A full stacktrace is below. This is happening with versions: xgboost.__version__ = 2.1.0

Fix

Currently the tutorial lists:
qid = rng.integers(0, n_query_groups, size=X.shape[0])
and should be
qid = sorted( rng.integers(0, n_query_groups, size=X.shape[0]) )

There is also a import pandas as pd added

Tests

# python
Python 3.10.14 (main, Jun 13 2024, 06:43:06) [GCC 12.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import sklearn
>>> sklearn.__version__
'1.5.0'
>>> import numpy as np
>>> np.__version__
'2.0.0'
>>> import xgboost as xgb
>>> xgb.__version__
'2.1.0'

# follow tutorial

>>> from sklearn.datasets import make_classification
>>> import numpy as np
>>> import xgboost as xgb
>>> # Make a synthetic ranking dataset for demonstration
>>> seed = 1994
>>> X, y = make_classification(random_state=seed)
>>> rng = np.random.default_rng(seed)
>>> n_query_groups = 3
>>> qid = rng.integers(0, n_query_groups, size=X.shape[0])
>>> # Sort the inputs based on query index
>>> sorted_idx = np.argsort(qid)
>>> X = X[sorted_idx, :]
>>> y = y[sorted_idx]
>>> ranker = xgb.XGBRanker(tree_method="hist", lambdarank_num_pair_per_sample=8, objective="rank:ndcg", lambdarank_pair_method="topk")
>>> ranker.fit(X, y, qid=qid)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 726, in inner_f
    return func(**kwargs)
  File "/usr/local/lib/python3.10/site-packages/xgboost/sklearn.py", line 1997, in fit
    train_dmatrix, evals = _wrap_evaluation_matrices(
  File "/usr/local/lib/python3.10/site-packages/xgboost/sklearn.py", line 596, in _wrap_evaluation_matrices
    train_dmatrix = create_dmatrix(
  File "/usr/local/lib/python3.10/site-packages/xgboost/sklearn.py", line 1879, in _create_ltr_dmatrix
    return super()._create_dmatrix(ref=ref, data=data, qid=qid, **kwargs)
  File "/usr/local/lib/python3.10/site-packages/xgboost/sklearn.py", line 1003, in _create_dmatrix
    return QuantileDMatrix(
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 726, in inner_f
    return func(**kwargs)
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 1573, in __init__
    self._init(
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 1632, in _init
    it.reraise()
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 569, in reraise
    raise exc  # pylint: disable=raising-bad-type
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 550, in _handle_exception
    return fn()
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 637, in <lambda>
    return self._handle_exception(lambda: self.next(input_data), 0)
  File "/usr/local/lib/python3.10/site-packages/xgboost/data.py", line 1416, in next
    input_data(**self.kwargs)
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 726, in inner_f
    return func(**kwargs)
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 626, in input_data
    self.proxy.set_info(
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 726, in inner_f
    return func(**kwargs)
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 962, in set_info
    self.set_uint_info("qid", qid)
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 1064, in set_uint_info
    dispatch_meta_backend(self, data, field, "uint32")
  File "/usr/local/lib/python3.10/site-packages/xgboost/data.py", line 1354, in dispatch_meta_backend
    _meta_from_numpy(data, name, dtype, handle)
  File "/usr/local/lib/python3.10/site-packages/xgboost/data.py", line 1295, in _meta_from_numpy
    _check_call(_LIB.XGDMatrixSetInfoFromInterface(handle, c_str(field), interface_str))
  File "/usr/local/lib/python3.10/site-packages/xgboost/core.py", line 284, in _check_call
    raise XGBoostError(py_str(_LIB.XGBGetLastError()))
xgboost.core.XGBoostError: [16:12:53] /workspace/src/data/data.cc:539: Check failed: non_dec: `qid` must be sorted in non-decreasing order along with data.
Stack trace:
  [bt] (0) /usr/local/lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x22d7cc) [0x7f3eb922d7cc]
  [bt] (1) /usr/local/lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x4b4468) [0x7f3eb94b4468]
  [bt] (2) /usr/local/lib/python3.10/site-packages/xgboost/lib/libxgboost.so(+0x4b4850) [0x7f3eb94b4850]
  [bt] (3) /usr/local/lib/python3.10/site-packages/xgboost/lib/libxgboost.so(XGDMatrixSetInfoFromInterface+0xb2) [0x7f3eb9134b42]
  [bt] (4) /lib/x86_64-linux-gnu/libffi.so.8(+0x6f7a) [0x7f3ef4093f7a]
  [bt] (5) /lib/x86_64-linux-gnu/libffi.so.8(+0x640e) [0x7f3ef409340e]
  [bt] (6) /lib/x86_64-linux-gnu/libffi.so.8(ffi_call+0xcd) [0x7f3ef4093b0d]
  [bt] (7) /usr/local/lib/python3.10/lib-dynload/_ctypes.cpython-310-x86_64-linux-gnu.so(+0xd66d) [0x7f3ef40a666d]
  [bt] (8) /usr/local/lib/python3.10/lib-dynload/_ctypes.cpython-310-x86_64-linux-gnu.so(+0xde1c) [0x7f3ef40a6e1c]

# qid is not sorted
>>> qid
array([1, 1, 2, 0, 1, 1, 2, 2, 2, 0, 1, 1, 2, 0, 1, 1, 2, 1, 0, 2, 1, 1,
       2, 0, 1, 2, 2, 2, 1, 0, 0, 1, 0, 1, 0, 0, 2, 1, 2, 2, 0, 1, 1, 1,
       2, 1, 1, 1, 0, 2, 2, 2, 2, 1, 0, 0, 2, 2, 0, 2, 0, 1, 2, 0, 2, 2,
       0, 0, 1, 2, 0, 2, 2, 1, 2, 1, 0, 1, 1, 2, 1, 1, 2, 1, 2, 0, 1, 2,
       1, 2, 1, 0, 2, 2, 0, 1, 2, 0, 2, 1])

###########
# try sorting qid with sorted()
###########
>>> qid = sorted(rng.integers(0, n_query_groups, size=X.shape[0]))
>>> qid
[np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(0), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(1), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2), np.int64(2)]
>>> sorted_idx = np.argsort(qid)
>>> X = X[sorted_idx, :]
>>> y = y[sorted_idx]
>>> ranker = xgb.XGBRanker(tree_method="hist", lambdarank_num_pair_per_sample=8, objective="rank:ndcg", lambdarank_pair_method="topk")
>>> ranker.fit(X, y, qid=qid)
XGBRanker(base_score=None, booster=None, callbacks=None, colsample_bylevel=None,
          colsample_bynode=None, colsample_bytree=None, device=None,
          early_stopping_rounds=None, enable_categorical=False,
          eval_metric=None, feature_types=None, gamma=None, grow_policy=None,
          importance_type=None, interaction_constraints=None,
          lambdarank_num_pair_per_sample=8, lambdarank_pair_method='topk',
          learning_rate=None, max_bin=None, max_cat_threshold=None,
          max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
          max_leaves=None, min_child_weight=None, missing=nan,
          monotone_constraints=None, multi_strategy=None, n_estimators=None,
          n_jobs=None, ...)
>>> 

[trivialfis]

Hi, the qid is sorted by the sorted_idx

[jpizagno]

Hi, the qid is sorted by the sorted_idx

Thank you for commenting. Why does the stacktrace occur then?

(It does look like X and y are sorted)

[trivialfis]

@jpizagno Did you include

xgboost/doc/tutorials/learning_to_rank.rst

Line 61 in 61ac8ee

qid = qid[sorted_idx]

in the example?

[trivialfis]

I see, this has been fixed in #9673 . We are still trying to get the document build for 2.1 working #10470 .

[jpizagno]

I see, this has been fixed in #9673 . We are still trying to get the document build for 2.1 working #10470 .

@trivialfis "aw shucks" , I was hoping to contribute. 😄 Yes, that would fix it. When the build is working, the docs would be correct.

There is also the import pandas as pd in this PR. Maybe one can assume that people who want to learn xgboost will also know pandas, but it can't hurt to add that. thoughts?

[hcho3]

The stable version now correctly shows the line qid = qid[sorted_idx]: https://xgboost.readthedocs.io/en/stable/tutorials/learning_to_rank.html