Python: Stack DataFrame based on similar string in column name-CodePudding

I have a DataFrame as follows:

data = [[np.nan,12,122],[np.nan,1230,1287],[123,101,812739],[1143,12301230,252],[234,342,4546],[2445,3453,3457],[7897,8657,5675],   [46,5675,453],[76,484,3735],   [363,93,4568],   [385,568,367],   [458,846,4847],   [574,45747,658468],   [57457,46534,4675]]
df1 = pd.DataFrame(data, index=['2022-01-01', '2022-01-02', '2022-01-03', '2022-01-04',
                       '2022-01-05', '2022-01-06', '2022-01-07', '2022-01-08',
                       '2022-01-09', '2022-01-10', '2022-01-11', '2022-01-12',
                       '2022-01-13', '2022-01-14'], 
          columns=['col_A', 'col_B', 'col_C'])
df1.index = pd.to_datetime(df1.index)

periodicity_dict = {'daily':'1D', 'weekly':'1W'}
df_columns = df1.columns

for key, value in periodicity_dict.items():
    for col in df_columns:
        df1[col '_rolling_' key]= np.nan
        for i in df1[col].index[df1[col].index >= df1[col].first_valid_index() pd.Timedelta(value)]:
            df1[col '_rolling_' key].loc[i] = (df1[col].loc[i] - df1[col].loc[i-pd.Timedelta(value)])/df1[col].loc[i-pd.Timedelta(value)]
df1 = df1.filter(regex='rolling')

I'd like to get something of the following form:

                     col_A_rolling  col_B_rolling   col_C_rolling    Type   timeframe  
quantile    
    0.01                 -0.317000         -0.234         -0.0443     pct      weekly
    0.03                 -0.314636         -0.022            ...      pct      weekly
    0.05                 ...                 ...             ...      ...             
    0.10                 ...                 ...
    0.01                 ...                 ...
    0.03                 ...                 ...
    0.05                 ...                 ...
    0.10                 -0.306364          -.530023                  pct       daily

(NOTE: the numbers in this DataFrame are hypothetical)

How can this be done using .stack() and .strp() functions? If possible.

CodePudding user response：

Here's what I came up with. There might be a more clever way; I'm not sure.

Also, I don't know what the Type column is for...

df2 = (
    df1
    .rename_axis('date')
    .rename_axis('col', axis=1)
    .stack()
    .rename('value')
    .reset_index()
)
df2['timeframe'] = df2['col'].map(lambda s: s.split('_')[-1])
df2['col'] = df2['col'].map(lambda s: '_'.join(s.split('_')[:-1]))

def get_quantiles(df, timeframe):
    return (
        df.query('timeframe == @timeframe')
          .pivot(None, 'col', 'value')
          .quantile([0.01, 0.03, 0.05, 0.1])
          .rename_axis('quantile')
          .rename_axis(None, axis=1)
          .assign(Type='pct', timeframe=timeframe)
    )

result = pd.concat((get_quantiles(df2, 'weekly'),
                    get_quantiles(df2, 'daily')))

Result:

          col_A_rolling  col_B_rolling  col_C_rolling Type timeframe
quantile
0.01          -0.761152      -0.976347      -0.945289  pct    weekly
0.03          -0.752987      -0.929133      -0.847109  pct    weekly
0.05          -0.744822      -0.881919      -0.748929  pct    weekly
0.10          -0.724408      -0.763884      -0.503479  pct    weekly
0.01          -0.974285      -0.990122      -0.998875  pct     daily
0.03          -0.934505      -0.970421      -0.997246  pct     daily
0.05          -0.894725      -0.950721      -0.995616  pct     daily
0.10          -0.795276      -0.917252      -0.978355  pct     daily