[英]Merge two dataframe if one string column is contained in another column in Pandas
[英]Pandas: merge two dataframe of different sizes on one column
我的第一個數據框 (df1) 如下所示:
pvalue trend time
0 0.000065 0.000076 2019-03-18 04:00:04
1 0.000087 0.000098 2019-03-18 04:00:06
2 0.000000 0.000000 2019-03-18 04:00:22
3 0.000000 0.000087 2019-03-18 04:02:29
4 0.000000 0.000000 2019-03-18 04:03:04
5 0.000000 0.000023 2019-03-18 04:03:05
6 0.000000 0.000000 2019-03-18 04:03:18
7 0.000000 0.000067 2019-03-18 04:18:55
8 0.000000 0.000000 2019-03-18 04:18:56
9 0.000000 0.000000 2019-03-18 04:20:41
我的第二個數據 (df2) 看起來像這樣:
time price
0 2019-03-18 04:00:00 0.00190633
1 2019-03-18 04:00:01 0.00190633
2 2019-03-18 04:00:02 0.00190633
3 2019-03-18 04:00:03 0.00190633
4 2019-03-18 04:00:04 0.00190633
5 2019-03-18 04:00:05 0.00190633
6 2019-03-18 04:00:06 0.00190800
7 2019-03-18 04:00:07 0.00190800
8 2019-03-18 04:00:08 0.00190800
9 2019-03-18 04:00:09 0.00190800
df2['time']
在每一行,有一秒的變化。 但是在 df1 上,每個df1['time']
之間有幾秒鍾的一些差距......我想要的是以下內容:
time price pvalue trend
0 2019-03-18 04:00:00 0.00190633 0.000000 0.000000
1 2019-03-18 04:00:01 0.00190633 0.000000 0.000000
2 2019-03-18 04:00:02 0.00190633 0.000000 0.000000
3 2019-03-18 04:00:03 0.00190633 0.000000 0.000000
4 2019-03-18 04:00:04 0.00190633 0.000065 0.000076
5 2019-03-18 04:00:05 0.00190633 0.000000 0.000000
6 2019-03-18 04:00:06 0.00190800 0.000087 0.000098
所以基本上擁有所有的秒數,當 df1 中有 pvalue 和趨勢的數據時,將它們放入新的數據框中。 我嘗試的是以下內容: df_all =df_pvalue_trade.merge(df_check,on='time',left_index=True)
但我只有 df1 的行,而不是像我的例子中的每一秒......知道嗎? 謝謝!
我使用上述嘗試過的代碼的結果如下:
pvalue trend time mkt_result price
6 0.000000 0.000000 2019-03-18 04:00:06 reject Ha := upward OR downward trend 0.00190800
21 0.000000 0.000000 2019-03-18 04:00:21 reject Ha := upward OR downward trend 0.00190800
22 0.000000 0.000000 2019-03-18 04:00:22 reject Ha := upward OR downward trend 0.00190800
149 0.000000 0.000000 2019-03-18 04:02:29 reject Ha := upward OR downward trend 0.00190594
184 0.000000 0.000000 2019-03-18 04:03:04 reject Ha := upward OR downward trend 0.00190594
185 0.000000 0.000000 2019-03-18 04:03:05 reject Ha := upward OR downward trend 0.00190594
198 0.000000 0.000000 2019-03-18 04:03:18 reject Ha := upward OR downward trend 0.00190594
這不是我想要的......
使用與DataFrame.fillna
merge
:
df = pd.merge(df2, df1, on='time', how='left').fillna(0)
print (df)
time price pvalue trend
0 2019-03-18 04:00:00 0.001906 0.000000 0.000000
1 2019-03-18 04:00:01 0.001906 0.000000 0.000000
2 2019-03-18 04:00:02 0.001906 0.000000 0.000000
3 2019-03-18 04:00:03 0.001906 0.000000 0.000000
4 2019-03-18 04:00:04 0.001906 0.000065 0.000076
5 2019-03-18 04:00:05 0.001906 0.000000 0.000000
6 2019-03-18 04:00:06 0.001908 0.000087 0.000098
7 2019-03-18 04:00:07 0.001908 0.000000 0.000000
8 2019-03-18 04:00:08 0.001908 0.000000 0.000000
9 2019-03-18 04:00:09 0.001908 0.000000 0.000000
此外,如果需要更換NaN
從唯一的列df1.columns
這就像不同df2.columns
:
d = dict.fromkeys(df1.columns.difference(df2.columns), 0)
print (d)
{'pvalue': 0, 'trend': 0}
df = pd.merge(df2, df1, on='time', how='left').fillna(d)
print (df)
time price pvalue trend
0 2019-03-18 04:00:00 0.001906 0.000000 0.000000
1 2019-03-18 04:00:01 0.001906 0.000000 0.000000
2 2019-03-18 04:00:02 0.001906 0.000000 0.000000
3 2019-03-18 04:00:03 0.001906 0.000000 0.000000
4 2019-03-18 04:00:04 0.001906 0.000065 0.000076
5 2019-03-18 04:00:05 0.001906 0.000000 0.000000
6 2019-03-18 04:00:06 0.001908 0.000087 0.000098
7 2019-03-18 04:00:07 0.001908 0.000000 0.000000
8 2019-03-18 04:00:08 0.001908 0.000000 0.000000
9 2019-03-18 04:00:09 0.001908 0.000000 0.000000
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.