在做线性回归模型时，这里提示字符串转换不了浮点数

import pandas as pd
import numpy as np
from sklearn import linear_model
from sklearn.model_selection import train_test_split
data=pd.read_csv('modelx.csv')
x_train,x_test=train_test_split(x,random_state=1)
x_test.shape
lr=linear_model.LinearRegression()
lr.fit(x_train,x_test)
print('Training score;{:2f}'.format(lr.score(x_train)))



ValueError: could not convert string to float: '好再见'

对文件里的符号和空格都进行处理了，但不知道为何还是发生这个错误，有人能帮我指点一下吗，感激不尽

lr.score(x_train))的值与预期不符，是字符串'好再见'