[英]Converting a pandas dataframe to a list of lists for input into an RNN
In Python, I have a dataframe that I imported with pandas.read_csv
that looks like this as an example:在 Python 中,我有一个用
pandas.read_csv
导入的数据pandas.read_csv
,例如如下所示:
Cust_id| time_to_event_f |event_id |event_sub_id
1 100 5 2
1 95 1 3
1 44 3 1
2 99 5 5
2 87 2 2
2 12 3 3
The data are ordered by cust_id
and then time_to_event_f
.数据按
cust_id
和time_to_event_f
。 I am trying to convert this dataframe into a tensor of dimensions [2,3,3]
so that for each customer id I have a sequential list of time_to_event_f
, event_id
, and event_sub_id
.我正在尝试将此数据帧转换为维度
[2,3,3]
的张量,以便对于每个客户 ID,我都有一个time_to_event_f
、 event_id
和event_sub_id
的顺序列表。 The idea is to use this as an input into an RNN in tensorflow.这个想法是将其用作张量流中 RNN 的输入。 I am following this tutorial so I am trying to get my data in a similar format.
我正在关注本教程,所以我试图以类似的格式获取我的数据。
You can transform the original dataframe d
to customer-id centered series by setting a Cust_id
index and then stacking:您可以通过设置
Cust_id
索引然后堆叠将原始数据帧d
转换为以客户 ID 为中心的系列:
d.set_index('Cust_id').stack()
The result series will look like this:结果系列将如下所示:
Cust_id
1 time_to_event_f 100
event_id 5
event_sub_id 2
time_to_event_f 95
event_id 1
event_sub_id 3
time_to_event_f 44
event_id 3
event_sub_id 1
2 time_to_event_f 99
event_id 5
event_sub_id 5
time_to_event_f 87
event_id 2
event_sub_id 2
time_to_event_f 12
event_id 3
event_sub_id 3
dtype: int64
Given this representation, you task is easy: take the values
ndarray and reshape it to your target size:鉴于这种表示,您的任务很简单:
values
ndarray 并将其重塑为您的目标大小:
series.values.reshape([2, 3, 3])
This array can be fed as input to tensorflow RNN.该数组可以作为 tensorflow RNN 的输入。 A complete code below:
完整代码如下:
import pandas as pd
from io import StringIO
s = StringIO("""
1 100 5 2
1 95 1 3
1 44 3 1
2 99 5 5
2 87 2 2
2 12 3 3
""".strip())
d = pd.read_table(s, names=['Cust_id', 'time_to_event_f', 'event_id', 'event_sub_id'], sep=r'\s+')
series = d.set_index('Cust_id').stack()
time_array = series.values.reshape([2, 3, 3])
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.