Feature importance (permutation or shapley values)

[bvphillips]

Other than the model importnaces given by RGF I would like to verify with permutation feature importance (scikit-learn) and/or with shapely values (shap package). But I can not do either, please help.

Environment Info

ACSVCPORT: 17532
ALLUSERSPROFILE: C:\ProgramData
APPDATA: C:\Users\bvphi\AppData\Roaming
ASL.LOG: Destination=file
COMMONPROGRAMFILES: C:\Program Files\Common Files
COMMONPROGRAMFILES(X86): C:\Program Files (x86)\Common Files
COMMONPROGRAMW6432: C:\Program Files\Common Files
COMPUTERNAME: LAPTOP-F77VAIJ2
COMSPEC: C:\Windows\system32\cmd.exe
CONDA_DEFAULT_ENV: spyMLenv
CONDA_EXE: C:\Users\bvphi\miniforge3\condabin..\Scripts\conda.exe
CONDA_EXES: "C:\Users\bvphi\miniforge3\condabin..\Scripts\conda.exe"
CONDA_PREFIX: C:\Users\bvphi\miniforge3\envs\spyMLenv
CONDA_PREFIX_1: C:\Users\bvphi\miniforge3
CONDA_PROMPT_MODIFIER: (spyMLenv)
CONDA_PYTHON_EXE: C:\Users\bvphi\miniforge3\python.exe
CONDA_SHLVL: 2
DRIVERDATA: C:\Windows\System32\Drivers\DriverData
HOMEDRIVE: C:
HOMEPATH: \Users\bvphi
LOCALAPPDATA: C:\Users\bvphi\AppData\Local
LOGONSERVER: \LAPTOP-F77VAIJ2
NUMBER_OF_PROCESSORS: 16
ONEDRIVE: C:\Users\bvphi\OneDrive
ONEDRIVECONSUMER: C:\Users\bvphi\OneDrive
OS: Windows_NT
PATH: C:\Users\bvphi\miniforge3\envs\spyMLenv;C:\Users\bvphi\miniforge3\envs\spyMLenv\Library\mingw-w64\bin;C:\Users\bvphi\miniforge3\envs\spyMLenv\Library\usr\bin;C:\Users\bvphi\miniforge3\envs\spyMLenv\Library\bin;C:\Users\bvphi\miniforge3\envs\spyMLenv\Scripts;C:\Users\bvphi\miniforge3\envs\spyMLenv\bin;C:\Users\bvphi\miniforge3\condabin;C:\Windows\system32;C:\Windows;C:\Windows\System32\Wbem;C:\Windows\System32\WindowsPowerShell\v1.0;C:\Windows\System32\OpenSSH;C:\Program Files (x86)\NVIDIA Corporation\PhysX\Common;C:\Program Files\NVIDIA Corporation\NVIDIA NvDLISR;C:\Program Files\Brandon Castellano\DVR-Scan;C:\Program Files\dotnet;C:\Users\bvphi\AppData\Local\Programs\Python\Python311\Scripts;C:\Users\bvphi\AppData\Local\Programs\Python\Python311;C:\Users\bvphi\AppData\Local\Microsoft\WindowsApps
PATHEXT: .COM;.EXE;.BAT;.CMD;.VBS;.VBE;.JS;.JSE;.WSF;.WSH;.MSC
PROCESSOR_ARCHITECTURE: AMD64
PROCESSOR_IDENTIFIER: AMD64 Family 25 Model 80 Stepping 0, AuthenticAMD
PROCESSOR_LEVEL: 25
PROCESSOR_REVISION: 5000
PROGRAMDATA: C:\ProgramData
PROGRAMFILES: C:\Program Files
PROGRAMFILES(X86): C:\Program Files (x86)
PROGRAMW6432: C:\Program Files
PROMPT: (spyMLenv) $P$G
PSMODULEPATH: C:\Program Files\WindowsPowerShell\Modules;C:\Windows\system32\WindowsPowerShell\v1.0\Modules
PUBLIC: C:\Users\Public
RLSSVCPORT: 22112
SYSTEMDRIVE: C:
SYSTEMROOT: C:\Windows
TEMP: C:\Users\bvphi\AppData\Local\Temp
TMP: C:\Users\bvphi\AppData\Local\Temp
USERDOMAIN: LAPTOP-F77VAIJ2
USERDOMAIN_ROAMINGPROFILE: LAPTOP-F77VAIJ2
USERNAME: bvphi
USERPROFILE: C:\Users\bvphi
WINDIR: C:\Windows
XML_CATALOG_FILES: file:https:///C:/Users/bvphi/miniforge3/envs/spyMLenv/etc/xml/catalog
LANG: en
SPYDER_ARGS: []
QT_SCALE_FACTOR:
QT_SCREEN_SCALE_FACTORS:
SPYDER_DEBUG_FILE: C:\Users\bvphi.spyder-py3\spyder-debug.log
SPY_EXTERNAL_INTERPRETER: False
SPY_UMR_ENABLED: True
SPY_UMR_VERBOSE: True
SPY_UMR_NAMELIST:
SPY_RUN_LINES_O:
SPY_PYLAB_O: True
SPY_BACKEND_O: 0
SPY_AUTOLOAD_PYLAB_O: False
SPY_FORMAT_O: 0
SPY_BBOX_INCHES_O: True
SPY_RESOLUTION_O: 72
SPY_WIDTH_O: 6
SPY_HEIGHT_O: 4
SPY_USE_FILE_O: False
SPY_RUN_FILE_O:
SPY_AUTOCALL_O: 0
SPY_GREEDY_O: False
SPY_JEDI_O: False
SPY_SYMPY_O: False
SPY_TESTING: False
SPY_HIDE_CMD: True
SPY_PYTHONPATH:
JPY_INTERRUPT_EVENT: 11012
IPY_INTERRUPT_EVENT: 11012
JPY_PARENT_PID: 11016
PYDEVD_USE_FRAME_EVAL: NO
TERM: xterm-color
CLICOLOR: 1
FORCE_COLOR: 1
CLICOLOR_FORCE: 1
PAGER: cat
GIT_PAGER: cat
MPLBACKEND: module:https://matplotlib_inline.backend_inline
KMP_INIT_AT_FORK: FALSE
KMP_DUPLICATE_LIB_OK: True

Operating System:
Microsoft Windows 10, x64-based PC
AMD Ryzen 7 5800HS Radeon Graphics, 3201 Mhz, 8 Cores, 16 Logical Processors

RGF/FastRGF/rgf_python version: 3.12.0

Python version (for rgf_python errors): 3.10.10

Error Message

runfile('C:/Users/bvphi/Documents/Python Scripts/new/rgf debug.py', wdir='C:/Users/bvphi/Documents/Python Scripts/new')
C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\rgf\utils.py:224: UserWarning: Cannot find FastRGF executable files. FastRGF estimators will be unavailable for usage.
warnings.warn("Cannot find FastRGF executable files. "
_RemoteTraceback:
"""
Traceback (most recent call last):
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\joblib\externals\loky\process_executor.py", line 428, in _process_worker
r = call_item()
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\joblib\externals\loky\process_executor.py", line 275, in call
return self.fn(*self.args, **self.kwargs)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\joblib_parallel_backends.py", line 620, in call
return self.func(*args, **kwargs)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\joblib\parallel.py", line 288, in call
return [func(*args, **kwargs)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\joblib\parallel.py", line 288, in
return [func(*args, **kwargs)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\utils\parallel.py", line 123, in call
return self.function(*args, **kwargs)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\inspection_permutation_importance.py", line 63, in _calculate_permutation_scores
scores.append(_weights_scorer(scorer, estimator, X_permuted, y, sample_weight))
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\inspection_permutation_importance.py", line 18, in _weights_scorer
return scorer(estimator, X, y)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\metrics_scorer.py", line 234, in call
return self._score(
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\metrics_scorer.py", line 276, in _score
y_pred = method_caller(estimator, "predict", X)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\metrics_scorer.py", line 73, in _cached_call
return getattr(estimator, method)(*args, **kwargs)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\rgf\utils.py", line 698, in predict
return self._estimators[0].predict(X)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\rgf\utils.py", line 352, in predict
self._execute_command(cmd)
File "C:\Users\bvphi\miniforge3\envs\spyMLenv\lib\site-packages\rgf\utils.py", line 303, in _execute_command
raise Exception(output)
Exception: "predict":
model_fn=C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.model-10
test_x_fn=C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.test.data.x
prediction_fn=C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.predictions.txt
Log:ON

Wed Jun 28 17:36:40 2023: Reading test data ...
!File I/O error!: (Detected in AzFile::seekReadBytes)
C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.test.data.x fread
"""
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File ~\miniforge3\envs\spyMLenv\lib\site-packages\spyder_kernels\py3compat.py:356 in compat_exec
exec(code, globals, locals)
File c:\users\bvphi\documents\python scripts\new\rgf debug.py:21
result = permutation_importance(do, X_test, y_test, scoring="r2",
File ~\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\inspection_permutation_importance.py:258 in permutation_importance
scores = Parallel(n_jobs=n_jobs)(
File ~\miniforge3\envs\spyMLenv\lib\site-packages\sklearn\utils\parallel.py:63 in call
return super().call(iterable_with_config)
File ~\miniforge3\envs\spyMLenv\lib\site-packages\joblib\parallel.py:1098 in call
self.retrieve()
File ~\miniforge3\envs\spyMLenv\lib\site-packages\joblib\parallel.py:975 in retrieve
self._output.extend(job.get(timeout=self.timeout))
File ~\miniforge3\envs\spyMLenv\lib\site-packages\joblib_parallel_backends.py:567 in wrap_future_result
return future.result(timeout=timeout)
File ~\miniforge3\envs\spyMLenv\lib\concurrent\futures_base.py:458 in result
return self.__get_result()
File ~\miniforge3\envs\spyMLenv\lib\concurrent\futures_base.py:403 in __get_result
raise self._exception
Exception: "predict":
model_fn=C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.model-10
test_x_fn=C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.test.data.x
prediction_fn=C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.predictions.txt
Log:ON

Wed Jun 28 17:36:40 2023: Reading test data ...
!File I/O error!: (Detected in AzFile::seekReadBytes)
C:\Users\bvphi\AppData\Local\Temp\rgf\31f6e18b-26f2-44d0-9b6f-aa37840498c31.test.data.x fread

<IPython.core.display.HTML object>
Traceback (most recent call last):
File ~\miniforge3\envs\spyMLenv\lib\site-packages\spyder_kernels\py3compat.py:356 in compat_exec
exec(code, globals, locals)
File c:\users\bvphi\documents\python scripts\new\rgf debug.py:36
ex = shap.TreeExplainer(rgf)
File ~\miniforge3\envs\spyMLenv\lib\site-packages\shap\explainers_tree.py:149 in init
self.model = TreeEnsemble(model, self.data, self.data_missing, model_output)
File ~\miniforge3\envs\spyMLenv\lib\site-packages\shap\explainers_tree.py:993 in init
raise InvalidModelError("Model type not yet supported by TreeExplainer: " + str(type(model)))
InvalidModelError: Model type not yet supported by TreeExplainer: <class 'rgf.rgf_model.RGFRegressor'>

Reproducible Example

Import modules

import gc
import time
import shap
import numpy as np
import pandas as pd
from rgf.sklearn import RGFRegressor
from rgf.utils import cleanup
import matplotlib.pyplot as plt
from sklearn.datasets import make_regression
from sklearn.model_selection import train_test_split
from sklearn.inspection import permutation_importance

rs = np.random.RandomState(2468097531)
X, y = make_regression(n_samples=15000, n_features=5)

Model

rgf = RGFRegressor()
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=rs)

PFI

start_time = time.time()
rgf.fit(X_train, y_train)
result = permutation_importance(rgf, X_test, y_test, scoring="r2",
n_repeats=10, n_jobs=-1, random_state=rs)
elapsed_time = time.time() - start_time
print(f"Elapsed time to compute the importances: {elapsed_time:.3f} seconds")
feature_names = ["1", "2", "3", "4", "5"]
Importances = pd.Series(result.importances_mean, index=feature_names)
fig, ax = plt.subplots()
Importances.plot.bar(yerr=result.importances_std, ax=ax)
ax.set_title("Permutations of full model")
ax.set_ylabel("Feature importances")
fig.tight_layout()
plt.show()
shap.initjs()
ex = shap.TreeExplainer(rgf)
shap_values = ex.shap_values(X_test)
shap.summary_plot(shap_values, X_test)
cleanup()
gc.collect()