Stuck on an issue?
Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.
And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.
Completely different output to original LGBMClassifier predict() and predict_proba() results
See original GitHub issueHi! I got completely different outputs for LGBMClassifier and “hummingbirded” model. It’s just tabular data with binary label.
Converted RandomForestClassifier works well on this data but it’s something wrong with lightgbm.
I tested hummingbird with lightgbm 3.2.1 and 2.1.1 version, results are the same. Outputs are completely different.
lgb_clf = lgb.LGBMClassifier()
lgb_clf.fit(X_train, y_train, eval_set=(X_test, y_test), verbose=10)
lgb_torch_model = convert(lgb_clf, 'pytorch')
lgb_clf.predict_proba(X_test)
>>
array([[0.0086527 , 0.9913473 ],
[0.08489025, 0.91510975],
[0.68027591, 0.31972409],
...,
[0.14574664, 0.85425336],
[0.35519048, 0.64480952],
[0.149126 , 0.850874 ]])
lgb_torch_model.predict_proba(X_test)
>>
array([[0.6863004 , 0.31369957],
[0.07888329, 0.9211167 ],
[0.6015551 , 0.39844486],
...,
[0.4556961 , 0.5443039 ],
[0.5472065 , 0.45279348],
[0.91947085, 0.08052915]], dtype=float32)
y_pred_lgb = lgb_clf.predict(X_test)
accuracy_score(y_test, y_pred_lgb)
>>
0.9557038067535095
y_pred_hmb = lgb_torch_model.predict(X_test)
accuracy_score(y_test, y_pred_hmb)
>>
0.6634469457442772
Python 3.9.5 OS: MacOS Big Sur 11.4
My environment:
# packages in environment at /opt/miniconda3/envs/hmb:
#
# Name Version Build Channel
anyio 3.2.1 py39h6e9494a_0 conda-forge
appnope 0.1.2 py39h6e9494a_1 conda-forge
argon2-cffi 20.1.0 py39hcbf5805_2 conda-forge
async_generator 1.10 py_0 conda-forge
attrs 21.2.0 pyhd8ed1ab_0 conda-forge
babel 2.9.1 pyh44b312d_0 conda-forge
backcall 0.2.0 pyh9f0ad1d_0 conda-forge
backports 1.0 py_2 conda-forge
backports.functools_lru_cache 1.6.4 pyhd8ed1ab_0 conda-forge
bleach 3.3.0 pyh44b312d_0 conda-forge
brotlipy 0.7.0 py39hcbf5805_1001 conda-forge
ca-certificates 2021.5.30 h033912b_0 conda-forge
certifi 2021.5.30 py39h6e9494a_0 conda-forge
cffi 1.14.5 py39h319c39b_0 conda-forge
chardet 4.0.0 py39h6e9494a_1 conda-forge
click 8.0.1 py39h6e9494a_0 conda-forge
cramjam 2.3.1 py39hce574ba_1 conda-forge
cryptography 3.4.7 py39ha2c9959_0 conda-forge
cycler 0.10.0 py_2 conda-forge
dbus 1.13.6 ha13b53f_2 conda-forge
decorator 5.0.9 pyhd8ed1ab_0 conda-forge
defusedxml 0.7.1 pyhd8ed1ab_0 conda-forge
dill 0.3.4 pyhd8ed1ab_0 conda-forge
entrypoints 0.3 pyhd8ed1ab_1003 conda-forge
expat 2.4.1 he49afe7_0 conda-forge
fastparquet 0.6.3 py39hc89836e_0 conda-forge
freetype 2.10.4 h4cff582_1 conda-forge
fsspec 2021.6.1 pyhd8ed1ab_0 conda-forge
future 0.18.2 py39h6e9494a_3 conda-forge
gensim 4.0.1 py39h9fcab8e_0 conda-forge
gettext 0.19.8.1 h7937167_1005 conda-forge
glib 2.68.3 he49afe7_0 conda-forge
glib-tools 2.68.3 he49afe7_0 conda-forge
hummingbird-ml 0.4.0 pyhd8ed1ab_0 conda-forge
icu 68.1 h74dc148_0 conda-forge
idna 2.10 pyh9f0ad1d_0 conda-forge
importlib-metadata 4.5.0 py39h6e9494a_0 conda-forge
ipykernel 5.5.5 py39h71a6800_0 conda-forge
ipython 7.24.1 py39h71a6800_0 conda-forge
ipython_genutils 0.2.0 py_1 conda-forge
ipywidgets 7.6.3 pyhd3deb0d_0 conda-forge
jbig 2.1 h0d85af4_2003 conda-forge
jedi 0.18.0 py39h6e9494a_2 conda-forge
jinja2 3.0.1 pyhd8ed1ab_0 conda-forge
joblib 1.0.1 pyhd8ed1ab_0 conda-forge
jpeg 9d hbcb3906_0 conda-forge
json5 0.9.5 pyh9f0ad1d_0 conda-forge
jsonschema 3.2.0 pyhd8ed1ab_3 conda-forge
jupyter 1.0.0 py39h6e9494a_6 conda-forge
jupyter_client 6.1.12 pyhd8ed1ab_0 conda-forge
jupyter_console 6.4.0 pyhd8ed1ab_0 conda-forge
jupyter_core 4.7.1 py39h6e9494a_0 conda-forge
jupyter_server 1.9.0 pyhd8ed1ab_0 conda-forge
jupyterlab 3.0.16 pyhd8ed1ab_0 conda-forge
jupyterlab_pygments 0.1.2 pyh9f0ad1d_0 conda-forge
jupyterlab_server 2.6.0 pyhd8ed1ab_0 conda-forge
jupyterlab_widgets 1.0.0 pyhd8ed1ab_1 conda-forge
kiwisolver 1.3.1 py39hedf5dff_1 conda-forge
krb5 1.19.1 hcfbf3a7_0 conda-forge
lcms2 2.12 h577c468_0 conda-forge
lerc 2.2.1 h046ec9c_0 conda-forge
libblas 3.9.0 8_mkl conda-forge
libcblas 3.9.0 8_mkl conda-forge
libclang 11.1.0 default_he082bbe_1 conda-forge
libcxx 11.1.0 habf9029_0 conda-forge
libdeflate 1.7 h35c211d_5 conda-forge
libedit 3.1.20191231 h0678c8f_2 conda-forge
libffi 3.3 h046ec9c_2 conda-forge
libgfortran 5.0.0 9_3_0_h6c81a4c_22 conda-forge
libgfortran5 9.3.0 h6c81a4c_22 conda-forge
libglib 2.68.3 hd556434_0 conda-forge
libiconv 1.16 haf1e3a3_0 conda-forge
liblapack 3.9.0 8_mkl conda-forge
libllvm11 11.1.0 hd011deb_2 conda-forge
libpng 1.6.37 h7cec526_2 conda-forge
libpq 13.3 hea3049e_0 conda-forge
libprotobuf 3.15.8 hcf210ce_0 conda-forge
libsodium 1.0.18 hbcb3906_1 conda-forge
libtiff 4.3.0 h1167814_1 conda-forge
libwebp-base 1.2.0 h0d85af4_2 conda-forge
lightgbm 3.2.1 pypi_0 pypi
llvm-openmp 11.1.0 hda6cdc1_1 conda-forge
lz4-c 1.9.3 h046ec9c_0 conda-forge
markupsafe 2.0.1 py39h89e85a6_0 conda-forge
matplotlib 3.4.2 py39h6e9494a_0 conda-forge
matplotlib-base 3.4.2 py39hb07454d_0 conda-forge
matplotlib-inline 0.1.2 pyhd8ed1ab_2 conda-forge
mistune 0.8.4 py39hcbf5805_1003 conda-forge
mkl 2020.4 h08c4f10_301 conda-forge
mysql-common 8.0.25 h694c41f_2 conda-forge
mysql-libs 8.0.25 h115446f_2 conda-forge
nbclassic 0.3.1 pyhd8ed1ab_1 conda-forge
nbclient 0.5.3 pyhd8ed1ab_0 conda-forge
nbconvert 6.1.0 py39h6e9494a_0 conda-forge
nbformat 5.1.3 pyhd8ed1ab_0 conda-forge
ncurses 6.2 h2e338ed_4 conda-forge
nest-asyncio 1.5.1 pyhd8ed1ab_0 conda-forge
ninja 1.10.2 h9a9d8cb_0 conda-forge
nltk 3.6.2 pyhd8ed1ab_0 conda-forge
notebook 6.4.0 pyha770c72_0 conda-forge
nspr 4.30 hcd9eead_0 conda-forge
nss 3.67 h31e2bf1_0 conda-forge
numpy 1.21.0 py39h7eed0ac_0 conda-forge
olefile 0.46 pyh9f0ad1d_1 conda-forge
onnx 1.9.0 py39hf443199_0 conda-forge
onnxconverter-common 1.8.1 pyhd8ed1ab_0 conda-forge
openjpeg 2.4.0 h6e7aa92_1 conda-forge
openssl 1.1.1k h0d85af4_0 conda-forge
packaging 20.9 pyh44b312d_0 conda-forge
pandas 1.2.5 py39h4d6be9b_0 conda-forge
pandoc 2.14.0.3 h0d85af4_0 conda-forge
pandocfilters 1.4.2 py_1 conda-forge
parso 0.8.2 pyhd8ed1ab_0 conda-forge
patsy 0.5.1 py_0 conda-forge
pcre 8.45 he49afe7_0 conda-forge
pexpect 4.8.0 pyh9f0ad1d_2 conda-forge
pickleshare 0.7.5 py_1003 conda-forge
pillow 8.2.0 py39h5fdd921_1 conda-forge
pip 21.1.2 pyhd8ed1ab_0 conda-forge
prometheus_client 0.11.0 pyhd8ed1ab_0 conda-forge
prompt-toolkit 3.0.19 pyha770c72_0 conda-forge
prompt_toolkit 3.0.19 hd8ed1ab_0 conda-forge
protobuf 3.15.8 py39h9fcab8e_0 conda-forge
psutil 5.8.0 py39h89e85a6_1 conda-forge
ptyprocess 0.7.0 pyhd3deb0d_0 conda-forge
pycparser 2.20 pyh9f0ad1d_2 conda-forge
pygments 2.9.0 pyhd8ed1ab_0 conda-forge
pyopenssl 20.0.1 pyhd8ed1ab_0 conda-forge
pyparsing 2.4.7 pyh9f0ad1d_0 conda-forge
pyqt 5.12.3 py39h6e9494a_7 conda-forge
pyqt-impl 5.12.3 py39hef7122c_7 conda-forge
pyqt5-sip 4.19.18 py39hd8f94c5_7 conda-forge
pyqtchart 5.12 py39hef7122c_7 conda-forge
pyqtwebengine 5.12.1 py39hef7122c_7 conda-forge
pyrsistent 0.17.3 py39hcbf5805_2 conda-forge
pysocks 1.7.1 py39h6e9494a_3 conda-forge
python 3.9.5 hd187cdc_0_cpython conda-forge
python-dateutil 2.8.1 py_0 conda-forge
python_abi 3.9 2_cp39 conda-forge
pytorch 1.8.0 cpu_py39h4f2e8f6_1 conda-forge
pytz 2021.1 pyhd8ed1ab_0 conda-forge
pyzmq 22.1.0 py39h7fec2f1_0 conda-forge
qt 5.12.9 h126340a_4 conda-forge
qtconsole 5.1.0 pyhd8ed1ab_0 conda-forge
qtpy 1.9.0 py_0 conda-forge
readline 8.1 h05e3726_0 conda-forge
regex 2021.4.4 py39h89e85a6_0 conda-forge
requests 2.25.1 pyhd3deb0d_0 conda-forge
requests-unixsocket 0.2.0 py_0 conda-forge
scikit-learn 0.24.2 py39h4b1dcc9_0 conda-forge
scipy 1.6.3 py39h056f1c0_0 conda-forge
seaborn 0.11.1 hd8ed1ab_1 conda-forge
seaborn-base 0.11.1 pyhd8ed1ab_1 conda-forge
send2trash 1.7.1 pyhd8ed1ab_0 conda-forge
setuptools 49.6.0 py39h6e9494a_3 conda-forge
six 1.16.0 pyh6c4a22f_0 conda-forge
sleef 3.5.1 h35c211d_1 conda-forge
smart_open 5.1.0 pyhd8ed1ab_1 conda-forge
sniffio 1.2.0 py39h6e9494a_1 conda-forge
sqlite 3.36.0 h23a322b_0 conda-forge
statsmodels 0.12.2 py39h329c335_0 conda-forge
terminado 0.10.1 py39h6e9494a_0 conda-forge
testpath 0.5.0 pyhd8ed1ab_0 conda-forge
threadpoolctl 2.1.0 pyh5ca1d4c_0 conda-forge
thrift 0.13.0 py39h9fcab8e_2 conda-forge
tk 8.6.10 h0419947_1 conda-forge
tornado 6.1 py39hcbf5805_1 conda-forge
tqdm 4.61.1 pyhd8ed1ab_0 conda-forge
traitlets 5.0.5 py_0 conda-forge
typing-extensions 3.10.0.0 hd8ed1ab_0 conda-forge
typing_extensions 3.10.0.0 pyha770c72_0 conda-forge
tzdata 2021a he74cb21_0 conda-forge
urllib3 1.26.5 pyhd8ed1ab_0 conda-forge
wcwidth 0.2.5 pyh9f0ad1d_2 conda-forge
webencodings 0.5.1 py_1 conda-forge
websocket-client 0.57.0 py39h6e9494a_4 conda-forge
wheel 0.36.2 pyhd3deb0d_0 conda-forge
widgetsnbextension 3.5.1 py39h6e9494a_4 conda-forge
xz 5.2.5 haf1e3a3_1 conda-forge
zeromq 4.3.4 h1c7c35f_0 conda-forge
zipp 3.4.1 pyhd8ed1ab_0 conda-forge
zlib 1.2.11 h7795811_1010 conda-forge
zstd 1.5.0 h582d3a0_0 conda-forge
Issue Analytics
- State:
- Created 2 years ago
- Comments:5
Top Results From Across the Web
How does the predict_proba() function in LightGBM work ...
First, the predict_proba() method of the class LGBMClassifier is calling the predict() method from LGBMModel (it inherits from it).
Read more >Different handling of similar looking data in LGBMClassifier ...
Look like if you use 1 * 4 numpy array, the problem remains: clf.predict_proba(np.array([list(X[1])])) . The issue is not checking predict input ...
Read more >lightgbm.LGBMClassifier — LightGBM 3.3.3.99 documentation
(for multi-class task) The predicted values. Predicted values are returned before any transformation, e.g. they are raw margin instead of ...
Read more >What is the difference between predict and predict_proba?
Hi! Consider a binary classification for labels 0 and 1. Predict will give either 0 or 1 as output. Predict_proba will give the...
Read more >Calibration of probabilities for tree-based models
Let us get validation predictions from a lightgbm model. ... Our model shows an AUC of 0.784, which is a nice result in...
Read more >
Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free
Top Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
I think no but I re-check it one more time and post results here
@interesaaat the reason of this behaviour was
intcolumns in my dataset. I just convert this columns tonp.float32and everything works fine. Thanks for helping me!