Question

当我试图预测z的值时，出现错误 “ ValueError：输入包含NaN，无穷大或对于dtype（'float32'）而言太大的值。”我在data.fillna(0, inplace=True)行中犯了错误还是其他问题？

import pandas as pd
import numpy as np
from sklearn import cross_validation
from sklearn.ensemble import RandomForestClassifier

data = pd.read_csv("C:/Users/Animkush/Desktop/train.csv")
data.replace("?", -99999, inplace=True)
data.drop(["Id"], 1, inplace=True)
data.convert_objects(convert_numeric=True)
data.fillna(0, inplace=True)

data1 = pd.read_csv("C:/Users/Animkush/Desktop/test.csv")
data1.replace("?", -99999, inplace=True)
data1.drop(["Id"], 1, inplace=True)
data.convert_objects(convert_numeric=True)
data.fillna(0, inplace=True)

def handle_non_numerical_data(data):
    columns = data.columns.values
    for column in columns:
            text_digit_vals = {}
            def convert_to_int(val):
            return text_digit_vals[val]

        if data[column].dtype != np.int64 and data[column].dtype != np.float64:
        column_contents = data[column].values.tolist()
        unique_elements = set(column_contents)
        x = 0
        for unique in unique_elements:
            if unique not in text_digit_vals:
                text_digit_vals[unique] = x
                x += 1

        data[column] = list(map(convert_to_int, data[column]))

return data

data = handle_non_numerical_data(data)
data1 = handle_non_numerical_data(data1)

x = np.array(data.drop(["SalePrice"], 1))
y = np.array(data["SalePrice"])
z = np.array(data1)

X_train, X_test, Y_train, Y_test = cross_validation.train_test_split(x, y,test_size=0.1)

clf = RandomForestClassifier()

clf.fit(X_train, Y_train)

print(clf.score(X_train, Y_train))

print(clf.predict(z))

Answer 1

您可能有inf值，请添加以下行

 data.replace([np.inf, -np.inf], 0, inplace=True)

无法从数据集中删除NaN值

1 个答案: