如何使用名义数据类型预测多项式回归 Python

问题描述

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

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：

多项式回归得分：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

我觉得我的预测列表很乱

解决方法

你搞砸了你的 pf.transform。在打印您的预测时 fit_transform，基本上您只在一个实例上适合您的转换，即您想要预测的那个。只需在您的训练集上进行 fit_transform，只需转换您的测试集并简单地转换您的预测列表。