combine multiple lines into one column

Here is my dataframe following the merging of two separate dataframes.

In the [order] column, orders 2, 4 and 5, have multiple corresponding amounts in the [value] column. This is replicated elsewhere in the dataframe at random points. What can I do to merge these amounts so that I am left with one row in the dataframe per order:

Essentially, how do I sum the values against each order line leaving me with one amount in the [value] columns for each row of the dataframe?

So my desired output will look something like:

order customer value  ordered     delivered
1     832      3450.0 2015-01-01  2015-01-02
2     803      607.0  2015-01-01  2015-01-02
3     105      135.0  2015-01-01  2015-01-02
4     57       667.8  2015-01-01  2015-01-02
5     905      1056.0 2015-01-01  2015-01-02

Answer

Use if order is column use GroupBy.transform with DataFrame.drop_duplicates:

df['value'] = df.groupby(df['order'].replace('', np.nan).ffill())['value'].transform('sum')

df = df.drop_duplicates('order')

If order is level of MultiIndex:

df['value'] = df.groupby(level=0)['value'].transform('sum')

df = df[~df.index.get_level_values('order').duplicated()]