"Input contains NaN, infinity or a value too large for dtype('float32')"当我训练决策树分类器

Question

我正试图为系外行星目录中的数据编写一个决策树方法。这是我硕士研究的一门课程。我在Jupyter笔记本上写过这个

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import sklearn
data = pd.read_csv('exoplanet.eu_catalog_2021.12.15.csv')
data_new = data.select_dtypes(include=['float64'])#Select only dtype float64 data
data_new[~data_new.isin([np.nan, np.inf, -np.inf]).any(1)]
data_new_2 = data_new.loc[:,('mass', 'mass_error_min')]
data_new_2.dropna(subset =["mass_error_min"], inplace = True)
data_new_2.info()
print(data_new_2)

结果是

<class 'pandas.core.frame.DataFrame'>
Int64Index: 1425 entries, 1 to 4892
Data columns (total 2 columns):
#   Column          Non-Null Count  Dtype  
---  ------          --------------  -----  
0   mass            1425 non-null   float64
1   mass_error_min  1425 non-null   float64
dtypes: float64(2)
memory usage: 33.4 KB

正如您所看到的，没有空单元格。此外，我写这篇文章是为了将所有数字转换为float64(以防万一！(

data_new_2['mass'] = data_new_2['mass'].astype(float)
data_new_2['mass_error_min'] = data_new_2['mass_error_min'].astype(float)

然后，我将数据拆分为训练和测试子集

from sklearn.model_selection import train_test_split
X = data_new_2.drop(["mass"], axis = 1)
y = data_new_2["mass"]
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = .30, random_state = 42)

没有问题。。。直到这部分

from sklearn.tree import DecisionTreeClassifier
classifier = DecisionTreeClassifier()
classifier.fit(X_train, y_train_2)

因为我收到这个错误消息

ValueError                                Traceback (most recent call last)
<ipython-input-327-7b81afce3234> in <module>
1 from sklearn.tree import DecisionTreeClassifier
2 classifier = DecisionTreeClassifier()
----> 3 classifier.fit(X_train, y_train_2)
.
.
.
~/.local/lib/python3.6/site-packages/sklearn/utils/validation.py in _assert_all_finite(X, allow_nan, msg_dtype)
104                     msg_err.format
105                     (type_err,
--> 106                      msg_dtype if msg_dtype is not None else X.dtype)
107             )
108     # for object dtype data, we only check for NaNs (GH-13254)
ValueError: Input contains NaN, infinity or a value too large for dtype('float32').

我不明白为什么会出现这个错误消息，因为我没有Nan、infitnity或"太大"；X_ train或y_ train数据中的数据。

我能做什么？

Answer 1

mass_error_min列中有一些无限值：

data_new_2.describe()
mass       mass_error_min
count   1425.000000       1425.0000
mean    6.060956          inf
std     13.568726         NaN
min     0.000002          0.0000
25%     0.054750          0.0116
50%     0.725000          0.0700
75%     3.213000          0.5300
max     135.300000        inf

所以，你必须用一些值来填充这些inf，使用以下代码：

value = data_new_2['mass_error_min'].quantile(0.98)
data_new_2 = data_new_2.replace(np.inf, value)