python pandas:整数列表作为DataFrame的单个值

时间:2015-10-28 19:31:56

标签: python list pandas

问题: 如何'pd.read_csv'以便给定列中的值为类型列表(列的每一行中的列表 )?

创建DataFrame时(从dict,见下文),各个值都是类型列表。问题:将DataFrame写入文件并从文件读回DataFrame后,我得到一个字符串而不是列表。

创建DataFrame
import pandas as pd
dict2df = {"euNOG": ["ENOG410IF52", "KOG2956", "KOG1997"], 
           "neg": [[58], [1332, 753, 716, 782], [187]], 
           "pos": [[96], [659, 661, 705, 1228], [1414]]}
df = pd.DataFrame(dict2df)

值是一个列表

type(df.loc[0, 'neg']) == list # --> True
type(df.loc[0, 'neg']) == str # --> False
df.loc[1, 'neg'][-1] == 782 # --> True
写入文件
df.to_csv('DataFrame.txt', sep='\t', header=True, index=False)
从文件中读取
df = pd.read_csv('DataFrame.txt', sep='\t')

value是一个不是列表的字符串

type(df.loc[0, 'neg']) == list # --> False
type(df.loc[0, 'neg']) == str # --> True
df.loc[1, 'neg'][-1] == 782 # --> False

当然,可以在两种数据类型之间进行转换,但它的计算成本很高,需要额外的工作(见下文)

def convert_StringList2ListOfInt(string2convert):
    return [int(ele) for ele in string2convert[1:-1].split(',')]

def DataFrame_StringOfInts2ListOfInts(df, cols2convert_list):
    for column in cols2convert_list:
        column_temp = column + "_temp"
        df[column_temp] = df[column].apply(convert_StringList2ListOfInt, 1)
        df[column] = df[column_temp]
        df = df.drop(column_temp, axis=1)
    return df
df = DataFrame_StringOfInts2ListOfInts(df, ['neg', 'pos'])

什么是更好(更pythonic)的解决方案?迭代列表中的整数而不必来回转换它们会非常方便。 谢谢你的支持!!

1 个答案:

答案 0 :(得分:2)

您可以使用ast.literal_eval()将字符串转换为列表。

ast.literal_eval() -

的简单示例
>>> import ast
>>> l = ast.literal_eval('[10,20,30]')
>>> type(l)
<class 'list'>

对于您的情况,您可以将其传递给Series.apply,以便(安全地)评估系列中的每个元素。示例 -

df = pd.read_csv('DataFrame.txt', sep='\t')
import ast
df['neg_list'] = df['neg'].apply(ast.literal_eval)
df = df.drop('neg',axis=1)
df['pos_list'] = df['pos'].apply(ast.literal_eval)
df = df.drop('pos',axis=1)

演示 -

In [15]: import pandas as pd

In [16]: dict2df = {"euNOG": ["ENOG410IF52", "KOG2956", "KOG1997"],
   ....:            "neg": [[58], [1332, 753, 716, 782], [187]],
   ....:            "pos": [[96], [659, 661, 705, 1228], [1414]]}

In [17]: df = pd.DataFrame(dict2df)

In [18]: df.to_csv('DataFrame.txt', sep='\t', header=True, index=False)

In [19]: newdf = pd.read_csv('DataFrame.txt', sep='\t')

In [20]: newdf['neg']
Out[20]:
0                     [58]
1    [1332, 753, 716, 782]
2                    [187]
Name: neg, dtype: object

In [21]: newdf['neg'][0]
Out[21]: '[58]'

In [22]: import ast

In [23]: newdf['neg_list'] = newdf['neg'].apply(ast.literal_eval)

In [24]: newdf = newdf.drop('neg',axis=1)

In [25]: newdf['pos_list'] = newdf['pos'].apply(ast.literal_eval)

In [26]: newdf = newdf.drop('pos',axis=1)

In [27]: newdf
Out[27]:
         euNOG               neg_list               pos_list
0  ENOG410IF52                   [58]                   [96]
1      KOG2956  [1332, 753, 716, 782]  [659, 661, 705, 1228]
2      KOG1997                  [187]                 [1414]

In [28]: newdf['neg_list'][0]
Out[28]: [58]