我是Python和Pandas的新手,我正在尝试将数组中的所有空值替换为特定值。
每次运行此命令时,更新的值都不会持久。
我已经看到Pandas在迭代行时不会保存更改...所以我如何保存更改?
这是我的代码
animal_kinds = set(df.AnimalKind) # this gives categories used below in the "ak" like dog, cat, bird
new_color_dog = 'polka dots'
new_color_cat = 'plaid'
new_color_bird = 'stripes'
for ak in animal_kinds:
ak_colors = ak['colors']
ak_with_no_color = animals[(df["Kind"] == ak ) & (df["Color"] == "" ) ]
result_count = len(ak_with_no_color)
if result_count:
ak_with_no_color.at["Color"] = new_color_ak #sets new color based on kind of animal (ak)
print(str(ak) 'color is changed to ' + str(new_color_ak))
答案 0 :(得分:1)
这种操作称为链接索引,它为 explicitly discouraged in the docs :
df[(df['kind'] == 'dog') & (df['colour'] == '')].at['colour'] = 'black'
相反,计算并使用布尔掩码:
mask = (df['kind'] == 'dog') & (df['colour'] == '')
df.loc[mask, 'colour'] = 'black'
这种操作在Python中不不起作用:
new_colour_dog = 'polka dots'
new_colour+'_dog' # want 'polka dots', but will not work
改为使用字典:
new_colours = {'dog': 'polka dots', 'cat': 'plaid', 'bird': 'stripes'}
然后您可以迭代字典的键/值对:
for animal, new_colour in new_colours.items():
mask = (df['kind'] == animal) & (df['colour'] == '')
df.loc[mask, 'colour'] = new_colour
当mask
返回一系列False
值时,您不需要测试/特殊情况的实例。