train() got an unexpected keyword argument 'reweighter'

[heury]

🐛 Bug Description

To Reproduce

Steps to reproduce the behavior:

1. python workflow.py get_data
2. python workflow.py dump_and_load_dataset
3. qrun .\examples\highfreq\workflow_config_High_Freq_Tree_Alpha158.yaml

Expected Behavior

I got an following error when running hifreq example.

TypeError: train() got an unexpected keyword argument 'reweighter'

I can fix it by adding reweighter parameter to highfreq_gdbt_model.py

def fit(
self,
dataset: DatasetH,
num_boost_round=1000,
early_stopping_rounds=50,
verbose_eval=20,
evals_result=dict(),
reweighter=None,
**kwargs
):

Is it a right solution to fix this error?

Screenshot

File "C:\dev\Anaconda3\envs\py38_torch\lib\site-packages\pyqlib-0.8.3.99-py3.8-win-amd64.egg\qlib\workflow\cli.py", line 57, in workflow
recorder = task_train(config.get("task"), experiment_name=experiment_name)
File "C:\dev\Anaconda3\envs\py38_torch\lib\site-packages\pyqlib-0.8.3.99-py3.8-win-amd64.egg\qlib\model\trainer.py", line 118, in task_train
_exe_task(task_config)
File "C:\dev\Anaconda3\envs\py38_torch\lib\site-packages\pyqlib-0.8.3.99-py3.8-win-amd64.egg\qlib\model\trainer.py", line 40, in exe_task
auto_filter_kwargs(model.fit)(dataset, reweighter=reweighter)
File "C:\dev\Anaconda3\envs\py38_torch\lib\site-packages\pyqlib-0.8.3.99-py3.8-win-amd64.egg\qlib\utils_init.py", line 852, in _func
return func(*args, **new_kwargs)
File "C:\dev\Anaconda3\envs\py38_torch\lib\site-packages\pyqlib-0.8.3.99-py3.8-win-amd64.egg\qlib\contrib\model\highfreq_gdbt_model.py", line 117, in fit
self.model = lgb.train(
TypeError: train() got an unexpected keyword argument 'reweighter'
PS C:\dev\qlib> python .\setup.py install

Environment

Note: User could run cd scripts && python collect_info.py all under project directory to get system information
and paste them here directly.

• Qlib version: pyqlib-0.8.3.99-py3.8-win-amd64.egg\
• Python version: 3.8
• OS (Windows, Linux, MacOS): Windows
• Commit number (optional, please provide it if you are using the dev version):

Additional Notes

[ly100120]

I 've ran into the same problem. Here is the source code for the reweighter from qlib.data.weight.

import pandas as pd
import numpy as np
from typing import Union, List, Tuple
from ...data.dataset import TSDataSampler
from ...data.dataset.utils import get_level_index
from ...utils import lazy_sort_index

class Reweighter:
def init(self, *args, **kwargs):
"""
To initialize the Reweighter, users should provide specific methods to let reweighter do the reweighting (such as sample-wise, rule-based).
"""
raise NotImplementedError()

def reweight(self, data: object) -> object:
    """
    Get weights for data

    Parameters
    ----------
    data : object
        The input data.
        The first dimension is the index of samples

    Returns
    -------
    object:
        the weights info for the data
    """
    raise NotImplementedError(f"This type of input is not supported")

so you can customize your own reweighter like this.
Good luck

[you-n-g]

@heury #908
Can you try this PR to checkout if the issue is fixed?

[heury]

It is working like charm. Thanks for your help.