Python convert value include into a dataframe as a dataframe-CodePudding

I'm still stuck on a transformation issue.

My goals is to convert rows included into a dataframe in a new dataframe

df0 = {'Table':['BKPF','BKPF','BKPF','BSEG','BSEG'],
'Zone':['TCODE', 'BLDAT', 'BLART','HKONT','GSBER'],
'Type':['CHAR','DATE','CHAR','CHAR','CHAR'],
'Len' :[10,8,2,10,4]}

into 2 new dataframes as

I'm new in Python, I have tried by different process like:

    df0 = pd.read_csv(Ztable, sep=';')
df1 = pd.DataFrame(df0)
t_bkpf = df1.loc[df1['Table'] == 'BBKPF']
t_bseg = df1.loc[df1['Table'] == 'BBSEG']

or with

t_bkpf = pd.DataFrame()
t_bseg = pd.DataFrame()
for index, row in df1.iterrows():
    if row['Table'] == 'BKPF':
        t_bkpf[row['Zone']] = np.nan

    if row['Table'] == 'BSEG':
        t_bseg[row['Zone']]= np.nan


df_bkpf = t_bkpf.copy()
df_bseg = t_bseg.copy()

The result based on code above it the best I have done but I have warning message on performance, I would like avoid it.

On next topic, I need to move data from one other Dataframe to these Dataframe based on columns defined.

Thanks

CodePudding user response：

You can do below:

df1 = pd.DataFrame(df0)
t = df1.set_index('Table')
dfs = [t.loc[idx].pivot(columns='Zone').stack().T for idx in t.index.unique()]

First dataframe:

print(dfs[0]):

Table  BKPF            
Zone  BLART BLDAT TCODE
Type   CHAR  DATE  CHAR
Len       2     8    10

Second dataframe:

print(dfs[1]):

Table  BSEG      
Zone  GSBER HKONT
Type   CHAR  CHAR
Len       4    10

CodePudding user response：

df = pd.DataFrame.from_dict(df0_dict, orient='index')
df

index	0	1	2	3	4
Table	BKPF	BKPF	BKPF	BSEG	BSEG
Zone	TCODE	BLDAT	BLART	HKONT	GSBER
Type	CHAR	DATE	CHAR	CHAR	CHAR
Len	10	8	2	10	4

rows = ['Zone', 'Type', 'Len']
df_BKPF = df.loc[rows, df.loc['Table'] == 'BKPF']
df_BKPF

index	0	1	2
Zone	TCODE	BLDAT	BLART
Type	CHAR	DATE	CHAR
Len	10	8	2

df_BSEG = df.loc[rows, df.loc['Table'] == 'BSEG']
df_BSEG

index	3	4
Zone	HKONT	GSBER
Type	CHAR	CHAR
Len	10	4