在处理熊猫数据框时遇到麻烦

Question

I occasionally am delivered a dataframe with many N/A values.

In these cases, there are reduntant rows. For every X value there is only one Y value. Therefore, I would like to merge the two "example1" rows into 1 row (as shown in the image), by combining the "context" column with measurement column names (M1,M2,..Mn).

How might one do this with pandas dataframe functions?

Thanks.

Answer 1

df = pd.DataFrame(
    [
        ['a', .1, np.nan, np.nan, .5],
        ['b', np.nan, .2, .3, .5],
    ],
    ['example1', 'example1'],
    ['context', 'M1', 'M2', 'M3', 'Y']
)

d1 = df.set_index('context', append=True).stack().unstack([1, 2])

d1.columns = d1.columns.map('_'.join)

d1

Answer 2

You could use a join. It takes in rsuffix and lsuffix parameters, so it would be easier to use those, but if you needed to use a prefix you could change it manually.

Create your DataFrame

df = pd.DataFrame({'X':['example1', 'example1'], 'context':['a',  'b'], 'M1':[0.1, np.nan], 'M2':[np.nan,0.2], 'M3':[np.nan, 0.3], 'Y':[0.5, 0.5]}, columns=['X', 'context', 'M1', 'M2', 'M3', 'Y'])

Solution

dfa = df[df['context'] == 'a'].set_index(['X', 'Y']).drop('context', axis=1)
dfb = df[df['context'] == 'b'].set_index(['X', 'Y']).drop('context', axis=1)

dfa.join(dfb, how='left', lsuffix='_a', rsuffix='_b').dropna(axis=1)