Hi I was removing some characters from the column names in my df but there is a lot of columns with diferent number of characters so I need to remove them specifically for diferent groups so I was using this code for the first 10 columns but it didn't work for the columns 19-29 because are replaced in the firsts columns:
new_names = dfr.iloc[:,0:10].rename(columns=lambda x: x[75:]).columns #first 10 columns
dfr.rename(columns=dict(zip(dfr.columns,new_names)),inplace=True)
The code that replace the first columns with the second group:
new_names = dfr.iloc[:,19:29].rename(columns=lambda x: x[38:]).columns #from 19-29 columns
dfr.rename(columns=dict(zip(dfr.columns,new_names)),inplace=True)
So I need to do this like for another 5 more groups some help to finish this.
Use:
np.random.seed(2020)
L = ['abdef','trasdfg','ssfgh','dfghj','jhgfdsa','kjhtf','ghrtd']
c = np.random.choice(L, size=10)
df = pd.DataFrame(np.random.randint(100, 105, size=(3, 10)), columns=c)
print (df)
abdef abdef dfghj ghrtd dfghj dfghj kjhtf dfghj abdef kjhtf
0 100 100 100 102 101 103 103 102 103 100
1 104 104 100 104 101 101 102 101 102 104
2 104 102 103 104 101 104 101 103 102 100
You can specify how many values are set to new columns names, here 3,4,3 and then in another list specify how many letters are filtered:
vals = [3,4,3]
filt=[2,3,1]
Then create tuples with cumulative sum and zip for start and end for filter:
cols = np.cumsum([0] + vals)
print (list( zip(zip(cols, cols[1:]), filt)))
[((0, 3), 2), ((3, 7), 3), ((7, 10), 1)]
In list comprehension create final columns names and set them back:
new = [x for (a,b),c in zip(zip(cols, cols[1:]), filt) for x in df.columns[a:b].str[:c]]
df.columns = new
print (df)
ab ab df ghr dfg dfg kjh d a k
0 100 100 100 102 101 103 103 102 103 100
1 104 104 100 104 101 101 102 101 102 104
2 104 102 103 104 101 104 101 103 102 100