Python Pandas规范化了大量的列名，优化了代码

Question

我正在制作一个脚本来处理我可以重用的csv。现在我正在使用这段代码来规范化csv文件中的列，以便它们都可以有类似的列。

df = pd.read_csv('Crokis.csv', index_col=0, encoding = "ISO-8859-1", low_memory=False)
genCol=['Genus','genus','ngenus','genera',]
df.rename(columns={typo: 'Genus' for typo in genCol}, inplace=True)
spCol=['species', 'sp', 'Species']
df.rename(columns={typo: 'species' for typo in spCol}, inplace=True)
chromCol=['Chromosome count', 'chromosome', 'Cytology', '2n', 'Chromosome']
df.rename(columns={typo: 'chromosome' for typo in chromCol}, inplace=True)
del chromCol, spCol, genCol

工作正常，但有两个问题

1. 有时由于大小写或在前面或后面添加了额外的字符，列表中缺少项目。是否有一种方法包括regex或类似的东西来处理不同的变化?

2. 似乎有一个多余的模式，所以我认为应该有一个方法来优化它。

Answer 1

可以使用python的re函数来实现。

下面是一个用'Genus'代替'genus.*'出现的例子。它将匹配和替换例如'genUS', 'GENUS', 'Genus_666'

import pandas as pd
import re
df = pd.read_csv('Crokis.csv', index_col=0, encoding = "ISO-8859-1", low_memory=False)
# 'Genus' column renaming
f = lambda x: re.sub('genus.*','Genus', x, flags = re.IGNORECASE)
df.rename(columns = f, inplace = True)

Answer 2

我将这样处理这个问题:

# use a single dict to hold the mapping
name_map = {'Genus': ['Genus','genus','ngenus','genera'],
        'species':['species', 'sp', 'Species'],
        'chromosome':['Chromosome count', 'chromosome', 'Cytology', '2n', 'Chromosome']}
col_translate = {}
for c in df.columns: 
    for canonical_name, alias_names in name_map.items():
        for alias_name in alias_names:
            if c.lower() == col_name.lower():
               col_translate[c] = canonical_name
            # if you want to check prefix or suffix...
            elif c.startswith(alias_name) or c.endswith(alias_name) 
               col_translate[c] = canonical_name
            # ... any additional, more complicated test
                ...

如果在某些情况下re可能认为太难