pandas join表的拼接操作

Relational Connection

merge

df1.merge(df2, on, how)
df.merge(dsf2, left_on, right_on, how)
df1.merge(df2, on, how, suffixes)

on is a column or a list, the base columns of merging.
Similarly, left_on for left table, right_on for right table.
how is a str in [‘left’, ‘right’, ‘inner’, ‘outer’], default ‘inner’
suffixes is a list with two suffixes,
used to distinguish columns with same name in different tables.

Specially,
merge can check out the correspondence in two tables
with a parameter validate

df1.merge(df2, on, how, validate)

validate is a str in [‘1:1’, ‘1:m’, ‘m:1’, ‘m:m’].

Index Connection

df1.join(df2, how, lsuffix, rsuffix)

Directional Connection

concat

pd.concat(dfs, axis, join, keys)

dfs is a list of dataframe joined.
axis is the direction, default 0.
join is a str in [‘left’, ‘right’, ‘inner’, ‘outer’], default ‘outer’.
keys is a list of tips to mark the origin of data.

append and assign

append adds a series to table as a new row.
assign adds a series to table as a new column.

df.append(s, ignore_index)
df.assign(col_name=s)

s is a series to be added.
ignore_index is a bool value.
If true, use a automatic index; if false, use s.name as index.
col_name is the name of new column.

Operation similar to connection

compare

df1.compare(df2)
df1.compare(df2, keep_shape=True)

It compares two tables, and outputs a dataframe of difference.
keep_shape means that keep the old shape, default False.

combine

df1.combine(df2, func)
df1.combine(df2, func, overwrite=False)

func,
inputs two column with same name in two tables,
outputs a new column.
overtwrite=False can preserve columns
in the called table
that do not appear in the passed in parameter table,
without setting missing values.