Add rows based on condition and also create and update new columns - pandas

Question

I have a pandas dataframe with few thousand rows, subset of it is below
fr var
1.1 10px
2.9 12pz
Expected Output:
fr var vard varv
1.1 10px -5 xval
1.1 10px 5 zval
2.9 12pz -6 zval
2.9 12pz 6 xval
For rows - Each row is to be split into two
Conditions for new columns:
'vard' - divide the numeric part of 'var' column by 2 and store it as two rows in 'vard', one negative and one positive value.
'varv' - if 'px' is in 'var column' and 'vard' has negative value, then 'varv' should be 'xval' else 'zval'.
Similarly if 'pz' is in 'var' column and 'vard' has negative value, then 'varv' should be 'zval' else 'xval'.
I have read through various answers with almost similar problems and tried many option like 'iterrows', 'shift', 'explode' etc but not able to get the expected output.

Answer 1

Use Series.str.extract first for numeric and non numeric part, convert numeric part to integers and divide by 2, then join with multiple value by -1 in concat, sorting index and create default, last use numpy.where for set new values by conditions:
df[['vard','varv']] = df['var'].str.extract('(\d+)(\D+)')
df['vard'] = df['vard'].astype(int).div(2)
df = pd.concat([df, df.assign(vard = df['vard']*-1)]).sort_index().reset_index(drop=True)
m = (df['varv'].eq('px') & df['vard'].lt(0)) | df['varv'].eq('pz') & df['vard'].gt(0)
df['varv'] = np.where(m, 'zval','xval')
print (df)
fr var vard varv
0 1.1 10px 5.0 xval
1 1.1 10px -5.0 zval
2 2.9 12pz 6.0 zval
3 2.9 12pz -6.0 xval

Answer 2

it is something that can easily be done using the melt function.
# recreate your dataframe
df = pd.DataFrame(columns=['fr','var'])
df['fr']=[1.1,2.9]
df['var']=['10px','12pz']
# split the var into its two components by creating two new columns
df['vard_p'] = df['var'].str[:-2]
df['vard_p'] = df['vard_p'].astype(float)/2
df['vard_n'] = -df['vard_p']
# get the vard from the var (I assumed it was simply the last character in the string)
df['varv'] = df['var'].str[-1]+'val'
# and here you melt on the two new vard columns to get the dataframe in the format you wanted
df = pd.melt(df, id_vars=['fr','var','varv'], value_vars=['vard_p','vard_n'])
# now rename or drop the new columns
df.rename(columns={'value':'vard'},inplace=True)
df.drop('variable',axis=1,inplace=True)
df
Output:
fr var varv vard
0 1.1 10px xval 5.0
1 2.9 12pz zval 6.0
2 1.1 10px xval -5.0
3 2.9 12pz zval -6.0
Hope it helped