我对numpy_pickle.py中的NumpyArrayWrapper进行了一些小的调整,以使决策树模型可以成功地加载到在z / OS上运行的scikit-learn中。更改归结为检查字节顺序是否正确以及是否不正确调用array.byteswap()。但是,当尝试加载GradientBoostingRegressor模型时,它甚至无法达到byteswap修复程序而失败。
错误来自此行https://github.com/scikit-learn/scikit-learn/blob/0.18.1/sklearn/tree/_tree.pyx#L644,这是由于以下条件node_ndarray.dtype != NODE_DTYPE
引起的。发生这种情况的原因是,当Boosting Regressor不会https://github.com/scikit-learn/scikit-learn/blob/0.18.1/sklearn/externals/joblib/numpy_pickle.py#L105
我想知道是否有人应该做些不同的事情,因为在z / OS上加载时,DT模型的Dtypes看起来不错,但是GBR模型却没有。这似乎来自model.fit方法,因为删除该调用时,我可以成功将pkl文件加载到z / OS上。
用于训练梯度提升模型的代码
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.pipeline import Pipeline
from sklearn import datasets
from sklearn import metrics
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import train_test_split
iris = datasets.load_iris()
X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target, test_size=0.2)
gbr = GradientBoostingRegressor(max_depth=3)
model = Pipeline([('Gbr', gbr)])
model.fit(X_train, y_train)
from sklearn.externals import joblib
joblib.dump(model, 'GBTmodelx86.pkl')
用于训练决策树模型的代码
from sklearn import datasets
from sklearn import metrics
from sklearn.tree import DecisionTreeClassifier
dataset = datasets.load_iris()
model = DecisionTreeClassifier()
model.fit(dataset.data, dataset.target)
from sklearn.externals import joblib
joblib.dump(model, 'DTmodelX86.pkl')
用于加载每个模型的代码
from sklearn.externals import joblib
model = joblib.load('DTmodelX86.pkl')
from sklearn.externals import joblib
model = joblib.load('GBTmodelx86.pkl')