保存多项式模型，不保存多项式次数答案

【问题标题】：saving polynomial model , doesn't save polynomial degree保存多项式模型，不保存多项式次数
【发布时间】：2020-11-12 23:18:56
【问题描述】：

当我想保存多项式模型时如何处理多项式次数，因为没有保存此信息！

import pandas as pd
import numpy as np
import joblib
from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
from sklearn.model_selection import train_test_split

df = pd.DataFrame({
        "a": np.random.uniform(0.0, 1.0, 1000),
        "b": np.random.uniform(10.0, 14.0, 1000),
        "c": np.random.uniform(100.0, 1000.0, 1000)})



def data():
    X_train, X_val, y_train, y_val  = train_test_split(df.iloc[:, :2].values,
                                                       df.iloc[:, 2].values,
                                                       test_size=0.2,
                                                       random_state=1340)
       
    return X_train, X_val, y_train, y_val


X_train, X_val, y_train, y_val = data()


poly_reg = PolynomialFeatures(degree = 2)
X_poly = poly_reg.fit_transform(X_train)

poly_reg_model = LinearRegression().fit(X_poly, y_train)



poly_model = joblib.dump(poly_reg_model, 'themodel')

y_pred = poly_reg_model.predict(poly_reg.fit_transform(X_val))

themodel = joblib.load('themodel')

现在，如果我尝试预测：

themodel.predict(X_val)，我正在接收：

ValueError: matmul: Input operand 1 has a mismatch in its core dimension 0, with gufunc signature (n?,k),(k,m?)->(n?,m?) (size 6 is different from 2)

我必须做的：

pol_feat = PolynomialFeatures(degree=2) themodel.predict(pol_feat.fit_transform(X_val))

为了工作。那么，我如何存储这些信息以便能够使用模型进行预测？

【问题讨论】：

标签： python-3.x machine-learning scikit-learn

【解决方案1】：

您还必须腌制训练有素的多项式特征：

# train and pickle
poly_reg = PolynomialFeatures(degree = 2)
X_poly = poly_reg.fit_transform(X_train)
poly_reg_model = LinearRegression().fit(X_poly, y_train)

joblib.dump(poly_reg_model, 'themodel')
joblib.dump(poly_reg, 'poilynomia_features_model')

# load and predict
poilynomia_features_model = joblib.load('poilynomia_features_model')
themodel = joblib.load('themodel')

X_val_prep = poilynomia_features_model.transform(X_val)
predictions = themodel.predict(X_val_prep)

但最好将所有步骤包含在单个 pipeline 中：

pipeline = Pipeline(steps=[('poilynomia', PolynomialFeatures()), 
                           ('lr', LinearRegression())])

pipeline.fit(X_train, y_train)
pipeline.predict(X_val)

【讨论】：