skranger-new 0.8.0

Python bindings for C++ ranger random forests

skranger provides scikit-learn compatible Python bindings to the C++ random forest implementation, ranger, using Cython.

The latest release of skranger uses version 0.12.1 of ranger.

Installation

skranger is available on pypi and can be installed via pip:

pip install skranger

If you encounter import errors with the wheel distribution, install directly from source:

pip install git+https://github.com/sacovo/skranger.git

Usage

There are two sklearn compatible classes, RangerForestClassifier and RangerForestRegressor. There is also the RangerForestSurvival class, which aims to be compatible with the scikit-survival API.

RangerForestClassifier

The RangerForestClassifier predictor uses ranger’s ForestProbability class to enable both predict and predict_proba methods.

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from skranger.ensemble import RangerForestClassifier

X, y = load_iris(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y)

rfc = RangerForestClassifier()
rfc.fit(X_train, y_train)

predictions = rfc.predict(X_test)
print(predictions)
# [1 2 0 0 0 0 1 2 1 1 2 2 2 1 1 0 1 1 0 1 1 1 0 2 1 0 0 1 2 2 0 1 2 2 0 2 0 0]

probabilities = rfc.predict_proba(X_test)
print(probabilities)
# [[0.01333333 0.98666667 0.        ]
#  [0.         0.         1.        ]
#  ...
#  [0.98746032 0.01253968 0.        ]
#  [0.99       0.01       0.        ]]

RangerForestRegressor

The RangerForestRegressor predictor uses ranger’s ForestRegression class. It also supports quantile regression using the predict_quantiles method.

from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split
from skranger.ensemble import RangerForestRegressor

X, y = load_boston(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y)

rfr = RangerForestRegressor()
rfr.fit(X_train, y_train)

predictions = rfr.predict(X_test)
print(predictions)
# [26.27401667  8.96549989 24.82981667 27.92506667 28.04606667 45.4693
#  21.89681787 40.30345    11.53959613 19.13675    15.88567273 16.69713567
#  ...
#  20.29025364 26.21245833 23.79643333 14.03546362 21.24893333 34.8825
#  21.22463333]

# enable quantile regression on instantiation
rfr = RangerForestRegressor(quantiles=True)
rfr.fit(X_train, y_train)

quantile_lower = rfr.predict_quantiles(X_test, quantiles=[0.1])
print(quantile_lower)
# [22.    5.   21.88 23.08 23.1  35.89 10.85 31.5   7.04 14.5  11.7  10.9
#   8.1  28.38  7.2  19.6  29.1  13.1  24.94 21.09 15.6  11.7  10.41 14.5
#  ...
#  18.9  21.4   9.43  8.7  26.46 18.99  7.2  19.27 18.5  21.19 18.99 18.88
#  14.07 21.87 22.18  9.43 17.28 29.6  18.2 ]
quantile_upper = rfr.predict_quantiles(X_test, quantiles=[0.9])
print(quantile_upper)
# [30.83 12.85 29.01 33.1  33.1  50.   29.75 50.   15.   23.   19.96 21.4
#  20.53 50.   13.35 25.   48.5  19.6  46.   26.6  23.7  20.1  17.8  21.4
#  ...
#  26.78 28.1  17.86 27.5  46.25 24.4  16.74 24.4  28.7  29.1  24.4  25.
#  25.   31.51 28.   20.8  26.7  42.13 24.24]

RangerForestSurvival

The RangerForestSurvival predictor uses ranger’s ForestSurvival class, and has an interface similar to the RandomSurvivalForest found in the scikit-survival package.

from sksurv.datasets import load_veterans_lung_cancer
from sklearn.model_selection import train_test_split
from skranger.ensemble import RangerForestSurvival

X, y = load_veterans_lung_cancer()
# select the numeric columns as features
X = X[["Age_in_years", "Karnofsky_score", "Months_from_Diagnosis"]]
X_train, X_test, y_train, y_test = train_test_split(X, y)

rfs = RangerForestSurvival()
rfs.fit(X_train, y_train)

predictions = rfs.predict(X_test)
print(predictions)
# [107.99634921  47.41235714  88.39933333  91.23566667  61.82104762
#   61.15052381  90.29888492  47.88706349  21.25111508  85.5768254
#   ...
#   56.85498016  53.98227381  48.88464683  95.58649206  48.9142619
#   57.68516667  71.96549206 101.79123016  58.95402381  98.36299206]

chf = rfs.predict_cumulative_hazard_function(X_test)
print(chf)
# [[0.04233333 0.0605     0.24305556 ... 1.6216627  1.6216627  1.6216627 ]
#  [0.00583333 0.00583333 0.00583333 ... 1.55410714 1.56410714 1.58410714]
#  ...
#  [0.12933333 0.14766667 0.14766667 ... 1.64342857 1.64342857 1.65342857]
#  [0.00983333 0.0112619  0.04815079 ... 1.79304365 1.79304365 1.79304365]]

survival = rfs.predict_survival_function(X_test)
print(survival)
# [[0.95855021 0.94129377 0.78422794 ... 0.19756993 0.19756993 0.19756993]
#  [0.99418365 0.99418365 0.99418365 ... 0.21137803 0.20927478 0.20513086]
#  ...
#  [0.87868102 0.86271864 0.86271864 ... 0.19331611 0.19331611 0.19139258]
#  [0.99021486 0.98880127 0.95299007 ... 0.16645277 0.16645277 0.16645277]]

License

skranger is licensed under GPLv3.

Development

To develop locally, it is recommended to have asdf, make and a C++ compiler already installed. After cloning, run make setup. This will setup the ranger submodule, install python and poetry from .tool-versions, install dependencies using poetry, copy the ranger source code into skranger, and then build and install skranger in the local virtualenv.

To format code, run make fmt. This will run isort and black against the .py files.

To run tests and inspect coverage, run make test.

To rebuild in place after making changes, run make build.

To create python package artifacts, run make dist.

To build and view documentation, run make docs.