pandas deform表的变形操作

Deform Table

pivot

pivot function depends on the uniqueness condition.

df.pivot(index, columns, values)

index is a column or list, as the index of new dataframe.
columns is a column or list, as the columns.
values is a column or list, as the values.

Simply,

More, index, columns, values are lists.

pivot_table

pivot_table does not depend on the uniqueness.
It’s aggfunc parameter appoints a aggregate function.\

df.pivot_table(index, columns, values, aggfunc, margins)

margins, bool, indicating whether perform marginal aggregation.
Other parameters are same as pivot's.

melt

Reverse process of pivot.

df.melt(id_vars, value_vars, var_name, value_name)

id_vars, a column or list, as the index of the new dataframe.
melt makes column names invalue_vars as a new column named var_name.
And it makes all origin values as a new column named value_name.

pd.wide_to_long

pd.wide_to_long(df,
                stubnames,
                i,
                j,
                sep,
                suffix)

i as index.
stubnames, sep, suffix, function uses RegEx to match these three parts.
It makes suffix as the values of a new index named j.
And it makes stubnames as columns.

Deform Index

df.unstack(levels)
df.stack(levels)

unstack transforms indexes in levels into columns.
Contrary to unstack, stack transforms columns into indexes.

Other

pd.crosstab

It is similar as to pivot_table

pd.croostab(index, columns, values, aggfunc, dropna)

dropna is a bool value, default True.
Do not include columns whose entries are all NaN.

explode

It could expand elements in a column, which is one of list, tuple, Series, np.ndarray.

df.explode(column)

get_dummies

It converts category features into indicator variables.

pd.get_dummies(df, dummy_na)

dummy_na is a bool value,
indicating whether regards NaN as a dummy.