python - How to change a value in a dataframe using multiple conditions with loc?

Question

I want to change a value from a row/column of a dataframe as follows:

Within the interval, when it found a value which is different equal to 1, then it should put 2.

The example:

initial df:

index                      event
2019-12-07 18:30:16         0
2019-12-07 19:30:16         0           
2019-12-07 20:30:16         0           
2019-12-07 21:30:16         0           
2019-12-07 22:30:16         1

wanted df:

index                      event
2019-12-07 18:30:16         0
2019-12-07 19:30:16         0           
2019-12-07 20:30:16         0           
2019-12-07 21:30:16         0           
2019-12-07 22:30:16         2

The following code works but I cannot change the value:

mask = (df.index > start_dates) & (df.index <= end_dates)

for k in range (0, len(df.loc[mask])):
    if df.loc[mask].event[k] == 1:
        df.loc[mask].loc[df.loc[mask].event == 1, "event"] = 2

I cannot change the value from 1 to 2 in the last line of code.

I also tried this...:

df.loc[mask].loc[df.loc[mask].event == 1, "event"] = 2
df.loc[mask].event[df.loc[mask].event == '1'] = 2
df.loc[mask].event[k] = 2

But none of the above lines works.

Please help me. :( Any help is highly appreciated!

Answer 1

The three lines you tried at the end are what pandas calls "chained loc calls", which will fail every time if you want to use them to assign new values. One .loc is enough for what you want

df.loc[mask] = 2
 # or
df.loc[mask, :] = 2
# both will assign two to all columns if you have more

# or also
df.loc[mask, 'event'] = 2

There is no need for the loop, .loc will select the rows you want with the boolean mask.

Edit

You can include a third condition to your mask

mask = (df.index > start_dates) & (df.index <= end_dates) & (df.event ==1)

Or leave your mask as it is and combine the conditions inside .loc

df.loc[mask & (df.event ==1), 'event'] = 2