Taking specific DataFrame row values and moving them to another new column

I have a DataFrame as follows:

Column1  Column2
'a'      'b'
'amount' '$d'
'e'      'f'
'amount' '$g' 
...

I created a new column 'Column3', where I would like to take the values of $d, $g, etc. (any value that is directly right to a value 'amount') and move them to Column 3, while also getting rid of value 'amount':

Column1  Column2  Column3
a        b        $d
e        f        $g
...

What would be the best way to move these values to Column3 and remove 'amount' values?

EDIT: Made the values more specific to the question.

Here's one approach:

• Create a boolean series with True where "Column1" equals "amount"(Series.eq).
• Apply Series.where to "Column2" to get values next to "amount", and use df.shift to shift the result one row up.
• Next, we add the series to the df with df.assign, and select from the df using the inverse (~) of the boolean series.
• Chain df.reset_index for a "clean" index.

m = df['Column1'].eq('amount')
res = df.assign(Column3=df['Column2'].where(m).shift(-1))[~m].reset_index(drop=True)

res

  Column1 Column2 Column3
0       a       b      $d
1       e       f      $g

With join/shift :

out = df[df["Column1"].ne("amount")].join(df["Column2"].rename("Column3").shift(-1))

​
Output :

print(out)

  Column1 Column2 Column3
0       a       b      $d
2       e       f      $g

Here is one way of doing it:

• Select values from Column2 that you want to move to a new column
• Create a new df, omitting rows where ["Column1"] != "c", and assign the pre-selected values to a new column
• (optional) reset index if you want to

Example:

# Pre-select values you want to move from `Column2`
data_to_move = df[df['Column1'] == 'c']['Column2'].values

# Drop rows where 'Column1' equals 'c', assigned pre-selected data to a new column
reshaped_data = df[df['Column1'] != 'c'].copy().reset_index(drop=True)
reshaped_data['Column3'] = data_to_move
reshaped_data

You can use a boolean mask:

m = df['Column1'] == 'amount'
df = df[~m].assign(Amount=df.loc[m, 'Column2'].values).reset_index(drop=True)

Output:

>>> df
  Column1 Column2 Amount
0       a       b     $d
1       e       f     $g