如何使用标称数据类型预测多项式回归 Python

Question

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score
from sklearn.preprocessing import PolynomialFeatures

df = pd.read_csv("diamonds.csv")

df = pd.get_dummies(df, columns = ["color", "clarity", "cut"])

X, Y = df.drop(labels = ["price", "color_E", "clarity_VS2", "cut_Good"], axis = 1).values, df[["price"]].values


pf = PolynomialFeatures(degree = 2, include_bias = False)
pf.fit(X_train)
pf.transform(X_train)

pf.transform(X_train)

X_train_transformed = pf.transform(X_train)
X_test_transformed = pf.transform(X_test)

modelR = LinearRegression()
modelR.fit(X_train_transformed, Y_train)


predictionlist = [0.23, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 61.5, 55, 3.47, 3.58, 1.57]

print("Polynomial Regression score: " + str(modelR.score(X_test_transformed, Y_test)) + " prediction: " + str(modelR.predict(pf.fit_transform([predictionlist]))[0][0]))

这是输出：

多项式回归分数：0.96599715147751 预测：-16308769.231718607

我的多项式回归的分数很好但我的预测很糟糕，钻石的价格怎么会是-16308769.231718607

我认为我的预测列表非常混乱

Answer 1

你搞砸了你的 pf.transform。当打印你的预测时 fit_transform，基本上你只在一个实例上拟合你的转换，你想要预测的那个。只需在您的训练集上进行 fit_transform 转换，只需转换您的测试集并转换您的预测列表。