sklearn 多元线性回归-> dtype 错误答案

【问题标题】：sklearn mutiple linear regression --> dtype errorsklearn 多元线性回归-> dtype 错误
【发布时间】：2021-01-12 04:56:56
【问题描述】：

我正在尝试使用线性回归模型来预测一个值。但是，当我使用 sklearn 中的 .predict 时，我找不到一种方法来插入 X 的数据而不会出现数据类型错误。

from sklearn import linear_model

KitchenQual_X = KitchenQual_df[["OverallQual", "YearBuilt", "YearRemodAdd", "GarageCars", "GarageArea"]]
KitchenQual_Y = KitchenQual_df["dummy_KitchenQual"]

regr_KitchenQual = linear_model.LinearRegression()
regr_KitchenQual.fit(KitchenQual_X, KitchenQual_Y)

print("Predicted missing KitchenQual value: " + regr_KitchenQual.predict(df_both[["OverallQual", "YearBuilt", "YearRemodAdd", "GarageCars", "GarageArea"]].loc[[1555]]))

在我的 kaggle 笔记本中运行代码时，我收到以下错误：

---------------------------------------------------------------------------
UFuncTypeError                            Traceback (most recent call last)
<ipython-input-206-1f022a48e21c> in <module>
----> 1 print("Predicted missing KitchenQual value: " + regr_KitchenQual.predict(df_both[["OverallQual", "YearBuilt", "YearRemodAdd", "GarageCars", "GarageArea"]].loc[[1555]]))

UFuncTypeError: ufunc 'add' did not contain a loop with signature matching types (dtype('<U37'), dtype('<U37')) -> dtype('<U37')

如果有任何帮助，我将不胜感激 :)

【问题讨论】：

您的数据中似乎有文本。
dtype('<U37') 是字符串，scikit 无法处理字符串。

标签： python pandas machine-learning scikit-learn linear-regression

【解决方案1】：

假设您的因变量是连续的，使用示例数据并重复您的步骤：

from sklearn import linear_model
import numpy as np
import pandas as pd

KitchenQual_df = pd.DataFrame(np.random.normal(0,1,(2000,6)))
KitchenQual_df.columns = ["OverallQual", "YearBuilt", "YearRemodAdd", "GarageCars", "GarageArea","dummy_KitchenQual"]

KitchenQual_X = KitchenQual_df[["OverallQual", "YearBuilt", "YearRemodAdd", "GarageCars", "GarageArea"]]
KitchenQual_Y = KitchenQual_df["dummy_KitchenQual"]

regr_KitchenQual = linear_model.LinearRegression()
regr_KitchenQual.fit(KitchenQual_X, KitchenQual_Y)

pred = regr_KitchenQual.predict(KitchenQual_df[["OverallQual", "YearBuilt", "YearRemodAdd", "GarageCars", "GarageArea"]].loc[[1555]])

预测是一个数组，你不能只使用+ 连接一个字符串和一个数组，下面的这些否定示例在问题中给你同样的错误：

"a" + np.array(['b','c'])
"a" + np.array([1,2])

UFuncTypeError: ufunc 'add' did not contain a loop with signature matching types (dtype('<U1'), dtype('<U1')) -> dtype('<U1')

你可以这样做：

print("Predicted missing KitchenQual value: " + str(pred[0]))

Predicted missing KitchenQual value: -0.11176904834490986

【讨论】：