I installed Conda with anaconda2 and Windows Subsystem for Linux.
I have strictly followed the steps in instructions, and tried the example named bench_ocrf,py
Here is the error information:
DeprecationWarning: This module has been deprecated in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.
"This module will be removed in 0.20.", DeprecationWarning)
/home/sy/anaconda/envs/OCRF_ENV/lib/python2.7/site-packages/sklearn/grid_search.py:43: DeprecationWarning: This module has been deprecated in favor of the model_selection module into which all the refactored classes and functions are moved. This module will be removed in 0.20.
DeprecationWarning)
OMP: Error #100: Fatal system error detected.
OMP: System error #22: Invalid argument
Aborted (core dumped)

Error compiling Cython file:

...

cdef inline DTYPE_t dist(self, DTYPE_t* x1, DTYPE_t* x2,
                         ITYPE_t size) except -1 with gil:
    cdef np.ndarray x1arr
    cdef np.ndarray x2arr
    with gil:
        ^

sklearn/neighbors/dist_metrics.pyx:1111:13: Trying to acquire the GIL while it is already held.
Traceback (most recent call last):
File "/home/rbrault/OCRF/build_tools/cythonize.py", line 198, in
main(root_dir_arg)
File "/home/rbrault/OCRF/build_tools/cythonize.py", line 190, in main
check_and_cythonize(root_dir)
File "/home/rbrault/OCRF/build_tools/cythonize.py", line 182, in check_and_cythonize
cythonize_if_unchanged(cur_dir, cython_file, gen_file, hashes)
File "/home/rbrault/OCRF/build_tools/cythonize.py", line 156, in cythonize_if_unchanged
cythonize(full_cython_path, full_gen_file_path)
File "/home/rbrault/OCRF/build_tools/cythonize.py", line 75, in cythonize
raise Exception('Cythonizing %s failed' % cython_file)
Exception: Cythonizing sklearn/neighbors/dist_metrics.pyx failed
Traceback (most recent call last):
File "setup.py", line 315, in
setup_package()
File "setup.py", line 299, in setup_package
generate_cython()
File "setup.py", line 185, in generate_cython
raise RuntimeError("Running cythonize failed!")
RuntimeError: Running cythonize failed!

Need investigation.

Linux-4.10.0-40-generic-x86_64-with-debian-stretch-sid
Python 3.6.3 |Anaconda custom (64-bit)| (default, Oct 16 2017, 15:28:36)
[GCC 4.8.2 20140120 (Red Hat 4.8.2-15)]
NumPy 1.13.3
SciPy 0.19.1
GCC 7.0.1

Related issue:
https://stackoverflow.com/questions/46920490/sklearn-installation-trying-to-acquire-the-gil-while-it-is-already-held

I am following your installation directions:

After running python setup.py install in python 2 environment I am getting this error:

Partial import of sklearn during the build process.
Generating cython files
Cythonizing sources
sklearn
sklearn/_isotonic.pyx has not changed
sklearn/__check_build/_check_build.pyx has not changed
sklearn/cluster/_dbscan_inner.pyx has not changed
Processing sklearn/cluster/_hierarchical.pyx

Error compiling Cython file:

...
from libcpp.map cimport map as cpp_map

Import the C-level symbols of numpy

cimport numpy as np

DTYPE = np.float64
^

sklearn/utils/fast_dict.pxd:13:10: cimported module has no attribute 'float64'
Traceback (most recent call last):
File "/home/javad/OCRF/build_tools/cythonize.py", line 198, in
main(root_dir_arg)
File "/home/javad/OCRF/build_tools/cythonize.py", line 190, in main
check_and_cythonize(root_dir)
File "/home/javad/OCRF/build_tools/cythonize.py", line 182, in check_and_cythonize
cythonize_if_unchanged(cur_dir, cython_file, gen_file, hashes)
File "/home/javad/OCRF/build_tools/cythonize.py", line 156, in cythonize_if_unchanged
cythonize(full_cython_path, full_gen_file_path)
File "/home/javad/OCRF/build_tools/cythonize.py", line 75, in cythonize
raise Exception('Cythonizing %s failed' % cython_file)
Exception: Cythonizing sklearn/cluster/_hierarchical.pyx failed
Traceback (most recent call last):
File "setup.py", line 315, in
setup_package()
File "setup.py", line 299, in setup_package
generate_cython()
File "setup.py", line 185, in generate_cython
raise RuntimeError("Running cythonize failed!")
RuntimeError: Running cythonize failed!
(py27) [javad@br005 OCRF]$

Hi :)

When I try to install this package following the instructions, I encounter the following error related to cythonizing.
Do you know the cause of the error?

Thanks,
Minori

Hi :)

I got this memory allocation error when running oneclassrf_error.py:

I changed the code a bit so that it doesn't GridSearchCV() but instead creates a model with OneClassRF() with default values for exercise.

Have any of you encounter this problem before and know how to fix the error?

• Aggarwal & Yu
• OCRF
• RF

Hi again :)

I would like to try out your One Class Random Forest Classifier (I read your paper as well).
However, in installing the code, I get some errors related to Cython files.

For example,

Do you know the problem or is there something I'm doing wrong?

Best,
Minori

Hello everybody,

I'm new to GitHub, so please excuse me if I've forgotten any of the usual information.

I'm following the EM_MV example (https://github.com/ngoix/EMMV_benchmarks). Since these methods use the one_class_data, I first wanted to install the OCRF.
I was able to run all the commands of the installation instructions without any problems - except for the last one: python setup.py install

Here I got the following error:

Cython is installed with the latest version:
(OCRF_env) Gerrits-Air:OCRF gerritschumann$ pip install --upgrade Cython

Result: Requirement already up-to-date: Cython in /Users/gerritschumann/anaconda3/envs/OCRF_env/lib/python2.7/site-packages (0.29.13)

To my knowledge, no other Cython versions are installed:
(OCRF_env) Gerrits-Air:OCRF gerritschumann$ python -c "import cython; print(cython.__version__)"

Result: 0.29.13

If relevant: I already had Python 3.6.9 installed on my Macbook. Due to the OCRF installation, I now have Python 2.7. as environment.

If there is no quick fix for the problem, is there any way to reconstruct the output of one_class_data so that I can play through the EM_MV example?

It would really help me a lot! Many Thanks in advance!

I have downloaded the code and installed the library with Conda following the instructions.
I tested both libraries of Isolation forest in this project (IsolationforestOCRF) and Isolation forest installed in the root environment of Conda (IsolationforestSklearn).

1. I found outputs of the method predict() in the two packages are different.
   The output of method predict() of IsolationforestOCRF is an vector with real numbers.
   The output of method predict() of IsolationforestSklearn is an vector with binary labels.
2. The result of anomaly detection with IsolationforestSklearn is much better than IsolationforestOCRF .
3. The results of OCRF and IsolationforestOCRF look similar.
4. The absolute value of output of method predict() and decision_function() of IsolationforestOCRF are
   same.
5. Running on the same dataset, however, the outputs of method decision_function() of
   IsolationforestOCRF and IsolationforestSklearn are quite different.
6. IsolationforestOCRF doesn't support parameter "contamination".

Regarding the issues listed above, is there any problem with the implementation of OCRF and IsolationforestOCRF in this package?

Hi,

My issue doesn't point out a specific problem, I just have one question about the OCRF unsupervised ROC and PR results (cf. bench_ocrf_roc_pr_unsupervised_factorized.png).

I presumed that this OCRF method refers to the "OCRFsampling" (Désir et al., 2013).

But I was wondering how these results generated with the bench_ocrf.py file, are supposed to be unsupervised since the fit_predict fonction needs the y_train and y_test
(cf line 97 : scoring = model.fit_predict(X_train, y_train, X_test, y_test)).

Thank you for your help.

Best regards,

MlleBerthe