How to convert repated rows data to columns in python? - python-3.x

Question

Hi I have a data frame df1 where column name names repeat after every 3 rows. I need to convert them to a single row.
This is how df looks
name marks
john 63
mark 45
pieter 32
beth 02
john 25
mark 01
pieter 23
beth 42
john 03
mark 43
pieter 42
beth 23
I need the output in the following format
type john mark pieter beth
marks 63 45 32 02
marks 25 01 23 42
marks 03 43 42 23

Answer 1

Consider this:
df=df.assign(id=df.groupby("name").cumcount()) \
.pivot(columns="name", index="id") \
.stack(level=0).reset_index(level=1) \
.rename(columns={"level_1": "type"})
del df.index.name
del df.columns.name
Outputs:
type beth john mark pieter
0 marks 02 63 45 32
1 marks 42 25 01 23
2 marks 23 03 43 42

Answer 2

IIUC:
new_df = (df.pivot_table(index=df.groupby('name').cumcount(), columns='name')
.rename_axis(columns=['type',None])
.stack(level=0)
.reset_index(level=1))
print(new_df)
type beth john mark pieter
0 marks 2 63 45 32
1 marks 42 25 1 23
2 marks 23 3 43 42
or
new_df = (df.assign(index=df.groupby('name').cumcount())
.melt(['index','name'], var_name='type')
.pivot_table(index=['type','index'], columns='name',values = 'value')
.reset_index('type'))

Answer 3

An alternative :
res = (df
.pivot(columns='name',values='marks')
.bfill()
#remove repititions on the beth column
#this impacts the other columns as well
.drop_duplicates('beth')
.rename_axis(columns=None)
.astype(int)
.assign(type='marks')
#adjust column positions to match ur output
.reindex(['type','john','mark','pieter','beth'],axis=1)
.reset_index(drop=True)
)
res
type john mark pieter beth
0 marks 63 45 32 2
1 marks 25 1 23 42
2 marks 3 43 42 23
You could also step out of Pandas into numpy, using the reshape method and at the end, create a new dataframe :
name_len = df.name.nunique()
names = df.name.unique()
df_len = len(df)
reshape_tuple = (df_len//name_len,name_len)
reshaped = df.marks.to_numpy().reshape(reshape_tuple)
#create new dataframe
res = pd.DataFrame(reshaped, columns = names)
#insert the 'type' column at the beginning of the dataframe
res.insert(0,'type','marks')
print(res)
type john mark pieter beth
0 marks 63 45 32 2
1 marks 25 1 23 42
2 marks 3 43 42 23