遍历不同的数据帧

Question

I have 59 different dataframes of different participants. Each participants has his/her own dataframe because that is necessary for splitting the right way in train and test groups later on.

Now I want to add a new column AcuteLegs_3days , which I knew works like this:

trainingload['AcuteLegs_3days']=trainingload["AcuteLegs_1day"].rolling(3).sum()
trainingload['AcuteLegs_3days']=trainingload['AcuteLegs_3days'].shift(1)

But can I do this for all 59 dataframes at once?

I have a list of all dataframe names:

participants = ['df_participant1', 'df_participant2', ...]

I tried to build a for loop but it is not working

for df in participants:
    df['AcuteLegs_3days']=df["AcuteLegs_1day"].rolling(3).sum()
    df['AcuteLegs_3days']=df['AcuteLegs_3days'].shift(1)

this gives the error:

TypeError: string indices must be integers

Answer 1

Your participants list is a list of strings, not a list of actual dataframes.

You can either create a list of dataframes, removing quotes:

participants = [df_participant1, df_participant2, ...]

and continue as you do now or get dataframe from locals (as I proposed in comments):

for df_name in participants: 
    df = locals()[df_name]
    df['AcuteLegs_3days']=df["AcuteLegs_1day"].rolling(3).sum()
    df['AcuteLegs_3days']=df['AcuteLegs_3days'].shift(1)

Currently you're trying to use strings as dataframes, which obviously fails (eg you have a list of names, not variables). So you try to execute the following:

'df_participant1'['AcuteLegs_3Days']

and strings can be indexed only by integers (well, strictly speaking, anything with __index__ defined) or slices. You cannot say 'foo'['bar'] , because it simply makes no sense.