即使熊猫df中的队列是int。它也像字符串一样或以怪异的方式排序。
n_frames video img
1 Videos482 1.jpg
10 Videos482 2g.jpg
11 Videos482 2d.jpg
2 Videos482 1q.jpg
1 Videos484 234.jpg
100 Videos484 34.jpg
我想要的是:请参见n_frames列。
n_frames video img
1 Videos482 1.jpg
2 Videos482 1t.jpg
10 Videos482 2g.jpg
11 Videos482 1q.jpg
1 Videos484 234.jpg
100 Videos484 34.jpg
[当我检查df.dtypes
时,我得到:n_frames int32
我的代码:
print(df.dtypes)
df = df.groupby('video')
df.apply(lambda _df: _df.sort_values(by=['n_frames']))
df.apply(lambda _df: _df.to_csv("rubbish.csv", index=False ))
所以基本上。我想对video进行分组,以保持n_frames的递增顺序。上层数据是原始形式的小形式。原始>
n_frames,trainTest,classes,vdo_noext,img
1,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 1.jpg
10,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 10.jpg
100,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 100.jpg
101,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 101.jpg
102,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 102.jpg
103,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 103.jpg
104,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 104.jpg
105,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 105.jpg
106,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 106.jpg
107,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 107.jpg
108,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 108.jpg
109,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 109.jpg
11,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 11.jpg
110,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 110.jpg
111,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 111.jpg
112,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 112.jpg
113,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 113.jpg
114,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 114.jpg
115,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 115.jpg
116,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 116.jpg
117,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 117.jpg
118,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 118.jpg
119,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 119.jpg
12,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 12.jpg
120,1,1,train/Normal/Normal_Videos482_x264,train/Normal/Normal_Videos482_x264@ 120.jpg
更新:n_frames列dtype 即使在转换为int之前,也从来没有字符串和整数。为什么整数以这种奇怪的方式排序?
很难说没有数据,但是我会像这样切换行:
df = df.groupby('video')
df['n_frames'] = df['n_frames'].astype(int)
df.apply(lambda _df: _df.sort_values(by=['n_frames']))
df.apply(lambda _df: _df.to_csv("rubbish.csv", index=False ))
换言之,在groupby
之后更改数据类型