how = ‘left’, ‘right’, ‘outer’, ‘inner’, ‘cross’

The former four are usually used. The concept of them are shown in the graph. Since cross join is difficult to be applied as the dataframe is big, it is seldom used.

University of Michigan - Ann Arbor

The frequently asked problems about pandas useful functions can be shared here.

Pandas useful functions

https://pandas.pydata.org/docs/index.html

Pandas documentation

pandas.read_csv(filepath, delimiter=None, names=None)

We usually use read_csv() to read a csv file and turn it into dataframe. It is a very useful function. Sometimes we face the problem that the file has not a header and the first line will be the column names of the dataframe. To solve this problem, you can add your expected column names to "names" in the form of names = \['a', 'b', 'c'\]. 

Some times there are rows with exceptional column numbers. For example, 4 are needed but this row have 5. Thus, some error will stop the program. To solve this, you can set \[error_bad_lines = False\] to drop those lines.

pd.read_csv()

pd.concat([s1, s2])

We should notice that the default if to concat by index, which means you will have more rows after this operation. If you want to concat by columns, set axis = 1.

pd.concat()

These are two function of pandas dataframe which makes people confuse sometimes. Actually there is no difference in use between these two functions!

pd.isna() & pd.isnull()

pd.merge()

unique() is often used to find the unique elements. It usually returns an array. 

unique().tolist() can store the elements into a list.

len(unique()) can find how many unique elements there are.

pd.unique()

There are three sorting methods in pandas. 

* By rows: pd.sort_index()
* By columns: pd.sort_index(axis = 1)
* By values: pd.sort_values(by = 'col')

To change the order, set "ascending = False"

Learn Before

Related