我有一个<data-table #toolChangeTable [columns]="columns" [data]="data" [loading]="loading" (lazyLoad)="onLazyLoad($event)" [lazy]="true"
[lazyLoadOnInit]="false" [pageSize]="pageOptions.size" [multiselect]="true" [paging] = "true" [totalRecords]="total"
defaultSortField="necLec" (edit)="updateToolChange($event, toolChangeTable)" (select)="onSelect($event, toolChangeTable)" (unSelect)="onUnSelect($event)">
<ng-container actionStart>
<button mat-button (click)="onMultiRowUpdateClick()" (keypress.enter)="onMultiRowUpdateClick()"
[disabled]="this.resourced || hasSelectedNone">Multi-Edit</button>
<button mat-button (click)="clearSelections()" (keypress.enter)="clearSelections()">Clear All</button>
<button mat-button (click)="onAddToolChangeClick()" [disabled]="this.resourced">Add Tool Change</button>
<button mat-button (click)="onDeleteToolChangeClick()" (keypress.enter)="onDeleteToolChangeClick()"
[disabled]="!hasSelectedSingle">Delete Tool Change</button>
<button mat-button [disabled]="!hasSelectedSingle" (click)="onEditAuthoritiesClick()"
(keypress.enter)="onEditAuthoritiesClick()">Edit Tool Change
Authorities</button>
</ng-container>
</data-table>
（test3），看起来像这样（DataFrame是date格式）
pd.datetime其中第一列import pandas as pd
data = {'date': ['1890-07-01 00:00:00', '1890-07-08 00:00:00', '1890-07-15 00:00:00', '1890-07-22 00:00:00', '1890-07-29 00:00:00'],
'date.1': ['1890-07-07', '1890-07-14', '1890-07-21', '1890-07-28', '1890-08-04'],
'mean_temp(℃)': [23.3, 23.9, 28.3, 26.1, 26.8],
'max_temp(℃)': [32.3, 33.2, 35.8, 33.3, 34.6],
'min_temp(℃)': [18.9, 17.0, 22.5, 22.0, 22.3]}
df = pd.DataFrame(data)
df.set_index('date', inplace=True)
date.1 mean_temp(℃) max_temp(℃) min_temp(℃)
date
1890-07-01 00:00:00 1890-07-07 23.3 32.3 18.9
1890-07-08 00:00:00 1890-07-14 23.9 33.2 17.0
1890-07-15 00:00:00 1890-07-21 28.3 35.8 22.5
1890-07-22 00:00:00 1890-07-28 26.1 33.3 22.0
1890-07-29 00:00:00 1890-08-04 26.8 34.6 22.3
是数据帧的索引。
我正在渲染新数据（第一列的date是rendered_date var，第三列是（pd.to_datetime是next_value_ var array([[28.330473]], dtype=float32)的另一列）。
rendered_date = render_date(last_day.index.date) # rendering new datetime object
rendered_date = pd.to_datetime(rendered_date, format='%Y/%m/%d') # making it for pandas
d = {'date':[rendered_date], 'mean_temp(℃)':[next_value_]}
new_df = pd.DataFrame(data=d) # making new dataframe
new_df = new_df.set_index("date") # setting the same index
fr = [test3, new_df] # concating new DF with existing df (test3)
result = pd.concat(fr)
使结果底部看起来像
....some values ....
2020-07-31 00:00:00 2020-08-06 28.7 35.0 23.9
[2020-08-07 00:00:00] NaT [[28.330473]] NaN NaN
这不是我想要的。
我只想在result数据帧（或test3都可接受）的末尾附加数据，保留形状和索引。我该如何设置相同的格式？
赞
....some values ....
2020-07-31 00:00:00 2020-08-06 28.7 35.0 23.9
2020-08-07 00:00:00 NaT 28.330473 NaN NaN
答案 0 :(得分：1)
编辑
您的代码对我来说很好：
data = {'date': ['1890-07-01 00:00:00', '1890-07-08 00:00:00', '1890-07-15 00:00:00', '1890-07-22 00:00:00', '1890-07-29 00:00:00'],
'date.1': ['1890-07-07', '1890-07-14', '1890-07-21', '1890-07-28', '1890-08-04'],
'mean_temp': [23.3, 23.9, 28.3, 26.1, 26.8],
'max_temp': [32.3, 33.2, 35.8, 33.3, 34.6],
'min_temp': [18.9, 17.0, 22.5, 22.0, 22.3]}
df = pd.DataFrame(data)
df.date = pd.to_datetime(df.date)
df.set_index('date', inplace=True)
rendered_date = pd.to_datetime('2020-08-07')
next_value_ = 28.330473
d = {'date': [rendered_date], 'mean_temp': [next_value_]}
df = pd.concat([df, pd.DataFrame(d).set_index('date')])
输出
date.1 mean_temp max_temp min_temp
date
1890-07-01 1890-07-07 23.300000 32.3 18.9
1890-07-08 1890-07-14 23.900000 33.2 17.0
1890-07-15 1890-07-21 28.300000 35.8 22.5
1890-07-22 1890-07-28 26.100000 33.3 22.0
1890-07-29 1890-08-04 26.800000 34.6 22.3
2020-08-07 NaN 28.330473 NaN NaN
添加行的一种更惯用的方法是
df.loc[rendered_date] = {'mean_temp': next_value_}
# # or
# df.loc[rendered_date] = [np.nan, next_value_, np.nan, np.nan]
# # or even
# df.loc[rendered_date, 'mean_temp'] = next_value_
所有选项的输出均相同
但是，如果可以获取所有新数据并将其作为批处理追加，则比一次串联一行要快。使用numpy / C实现，使用Python更快地将标量值分配给本地对象（如列表或字典），而使用矢量化调用的熊猫则更快。