使用来自另一个数据框的值更新熊猫数据框

Question

Assuming a dataframe where values from any of the columns can change, Given another dataframe which contains the old value, new value and column it belongs to, how to update dataframe using information about changes? For example:

>>> my_df
    x    y    z
0   1    2    5
1   2    3    9
2   8    7    2
3   3    4    7
4   6    7    7

my_df_2 contains information about changed values and their columns:

>>> my_df_2
    changed_col    old_value    new_value
0      x              2             10
1      z              9             20
2      x              1             12
3      y              4             23

How to use information in my_df_2 to update my_df such that my_df now becomes:

>>> my_df
    x     y     z
0   12    2     5
1   10    3     20
2   8     7     2
3   3     23    7
4   6     7     7

Answer 1

You can create a dictionary for the changes as follows:

d = {i: dict(zip(j['old_value'], j['new_value'])) for i, j in my_df_2.groupby('changed_col')}

d
Out: {'x': {1: 12, 2: 10}, 'y': {4: 23}, 'z': {9: 20}}

Then use it in DataFrame.replace :

my_df.replace(d)
Out: 
    x   y   z
0  12   2   5
1  10   3  20
2   8   7   2
3   3  23   7
4   6   7   7

Answer 2

You can use the update method. See http://pandas.pydata.org/pandas-docs/version/0.17.1/generated/pandas.DataFrame.update.html

Example:

old_df = pd.DataFrame({"a":np.arange(5), "b": np.arange(4,9)})

+----+-----+-----+
|    |   a |   b |
|----+-----+-----|
|  0 |   0 |   4 |
|  1 |   1 |   5 |
|  2 |   2 |   6 |
|  3 |   3 |   7 |
|  4 |   4 |   8 |
+----+-----+-----+

new_df = pd.DataFrame({"a":np.arange(7,8), "b": np.arange(10,11)})
+----+-----+-----+
|    |   a |   b |
|----+-----+-----|
|  0 |   7 |  10 |
+----+-----+-----+

old_df.update(new_df) 
+----+-----+-----+
|    |   a |   b |
|----+-----+-----|
|  0 |   7 |  10 | #Changed row
|  1 |   1 |   5 |
|  2 |   2 |   6 |
|  3 |   3 |   7 |
|  4 |   4 |   8 |
+----+-----+-----+