Pandas DataFrame Methods: drop_level(), pivot(), pivot_table(), reorder_levels(), sort_values() and sort_index()

http://img.youtube.com/vi/PMKuZoQoYE0/0.jpg

The Pandas DataFrame/Series has several methods to handle Missing Data. When applied to a DataFrame/Series, these methods evaluate and modify the missing elements.

This is Part 13 of the DataFrame methods series:

  • Part 1 focuses on the DataFrame methods abs(), all(), any(), clip(), corr(), and corrwith().
  • Part 2 focuses on the DataFrame methods count(), cov(), cummax(), cummin(), cumprod(), cumsum().
  • Part 3 focuses on the DataFrame methods describe(), diff(), eval(), kurtosis().
  • Part 4 focuses on the DataFrame methods mad(), min(), max(), mean(), median(), and mode().
  • Part 5 focuses on the DataFrame methods pct_change(), quantile(), rank(), round(), prod(), and product().
  • Part 6 focuses on the DataFrame methods add_prefix(), add_suffix(), and align().
  • Part 7 focuses on the DataFrame methods at_time(), between_time(), drop(), drop_duplicates() and duplicated().
  • Part 8 focuses on the DataFrame methods equals(), filter(), first(), last(), head(), and tail()
  • Part 9 focuses on the DataFrame methods equals(), filter(), first(), last(), head(), and tail()
  • Part 10 focuses on the DataFrame methods reset_index(), sample(), set_axis(), set_index(), take(), and truncate()
  • Part 11 focuses on the DataFrame methods backfill(), bfill(), fillna(), dropna(), and interpolate()
  • Part 12 focuses on the DataFrame methods isna(), isnull(), notna(), notnull(), pad() and replace()
  • Part 13 focuses on the DataFrame methods drop_level(), pivot(), pivot_table(), reorder_levels(), sort_values() and sort_index()

Getting Started

Remember to add the Required Starter Code to the top of each code snippet. This snippet will allow the code in this article to run error-free.

Required Starter Code

import pandas as pd
import numpy as np 

Before any data manipulation can occur, two new libraries will require installation.

  • The pandas library enables access to/from a DataFrame.
  • The numpy library supports multi-dimensional arrays and matrices in addition to a collection of mathematical functions.

To install these libraries, navigate to an IDE terminal. At the command prompt ($), execute the code below. For the terminal used in this example, the command prompt is a dollar sign ($). Your terminal prompt may be different.

$ pip install pandas

Hit the <Enter> key on the keyboard to start the installation process.

$ pip install numpy

Hit the <Enter> key on the keyboard to start the installation process.

Feel free to check out the correct ways of installing those libraries here:

If the installations were successful, a message displays in the terminal indicating the same.

DataFrame drop_level()

The drop_level() method removes the specified index or column from a DataFrame/Series. This method returns a DataFrame/Series with the said level/column removed.

The syntax for this method is as follows:

DataFrame.droplevel(level, axis=0)
Parameter Description
level If the level is a string, this level must exist. If a list, the elements must exist and be a level name/position of the index.
axis If zero (0) or index is selected, apply to each column. Default is 0 (column). If zero (1) or columns, apply to each row.

For this example, we generate random stock prices and then drop (remove) level Stock-B from the DataFrame.

nums = np.random.uniform(low=0.5, high=13.3, size=(3,4))
df_stocks = pd.DataFrame(nums).set_index([0, 1]).rename_axis(['Stock-A', 'Stock-B'])
print(df_stocks)

result = df_stocks.droplevel('Stock-B')
print(result)
  • Line [1] generates random numbers for three (3) lists within the specified range. Each list contains four (4) elements (size=3,4). The output saves to nums.
  • Line [2] creates a DataFrame, sets the index, and renames the axis. This output saves to df_stocks.
  • Line [3] outputs the DataFrame to the terminal.
  • Line [4] drops (removes) Stock-B from the DataFrame and saves it to the result variable.
  • Line [5] outputs the result to the terminal.

Output:

df_stocks

    2 3
Stock-A Stock-B    
12.327710 10.862572   7.105198  8.295885
11.474872 1.563040    5.915501  6.102915

result

  2 3
Stock-A    
12.327710 7.105198  8.295885
11.474872 5.915501  6.102915

DataFrame pivot()

The pivot() method reshapes a DataFrame/Series and produces/returns a pivot table based on column values.

The syntax for this method is as follows:

DataFrame.pivot(index=None, columns=None, values=None)
Parameter Description
index This parameter can be a string, object, or a list of strings and is optional. This option makes up the new DataFrame/Series index. If None, the existing index is selected.
columns This parameter can be a string, object, or a list of strings and is optional. Makes up the new DataFrame/Series column(s).
values This parameter can be a string, object, or a list of the previous and is optional.

For this example, we generate 3-day sample stock prices for Rivers Clothing. The column headings display the following characters.

  • A (for Opening Price)
  • B (for Midday Price)
  • C (for Opening Price)
cdate_idx = ['01/15/2022', '01/16/2022', '01/17/2022'] * 3
group_lst = list('AAABBBCCC')
vals_lst  = np.random.uniform(low=0.5, high=13.3, size=(9))

df = pd.DataFrame({'dates':  cdate_idx,
                                    'group':  group_lst,
                                   'value':  vals_lst})
print(df)

result = df.pivot(index='dates', columns='group', values='value')
print(result)
  • Line [1] creates a list of dates and multiplies this by three (3). The output is three (3) entries for each date. This output saves to cdate_idx.
  • Line [2] creates a list of headings for the columns (see above for definitions). Three (3) of each character are required (9 characters). This output saves to group_lst.
  • Line [3] uses np.random.uniform to create a random list of nine (9) numbers between the set range. The output saves to vals_lst.
  • Line [4] creates a DataFrame using all the variables created on lines [1-3]. The output saves to df.
  • Line [5] outputs the DataFrame to the terminal.
  • Line [6] creates a pivot from the DataFrame and groups the data by dates. The output saves to result.
  • Line [7] outputs the result to the terminal.

Output:

df

  dates group value
0 01/15/2022 A 9.627767
1 01/16/2022     A 11.528057
2 01/17/2022     A 13.296501
3 01/15/2022 B 2.933748
4 01/16/2022     B 2.236752
5 01/17/2022     B 7.652414
6 01/15/2022 C 11.813549
7 01/16/2022     C 11.015920
8 01/17/2022     C 0.527554

result

group A B C
dates      
01/15/2022   8.051752  9.571285   6.196394
01/16/2022  6.511448  8.158878  12.865944
01/17/2022  8.421245  1.746941  12.896975

DataFrame pivot_table()

The pivot_table() method streamlines a DataFrame to contain only specific data (columns). For example, say we have a list of countries with associated details. We only want to display one or two columns. This method can accomplish this task.

The syntax for this method is as follows:

DataFrame.pivot_table(values=None, index=None, columns=None, aggfunc='mean', fill_value=None, margins=False, dropna=True, margins_name='All', observed=False, sort=True)
Parameter Description
values This parameter is the column to aggregate and is optional.
index If the parameter is an array, it must be the same length as the data. It may contain any other data types (but not a list).
columns If an array, it must be the same length as the data. It may contain any other data types (but not a list).
aggfunc This parameter can be a list of functions. These name(s) will display at the top of the relevant column names (see Example 2).
fill_value This parameter is the value used to replace missing values in the table after the aggregation has occurred.
margins If set to True, this parameter will add the row/column data to create subtotal(s) or total(s). False, by default.
dropna This parameter will not include any columns where the value(s) are NaN. True by default.
margins_name This parameter is the name of the row/column containing the totals if margins parameter is True.
observed If True, display observed values. If False, display all observed values.
sort By default, sort is True. The values automatically sort. If False, no sort is applied.

For this example, a comma-delimited CSV file is read in. A pivot table is created based on selected parameters.

Code – Example 1:

df = pd.read_csv('countries.csv')
df = df.head(5)
print(df)

result = pd.pivot_table(df, values='Population', columns='Capital')
print(result)
  • Line [1] reads in a CSV file and saves to a DataFrame (df).
  • Line [2] saves the first five (5) rows of the CSV file to df (over-writing df).
  • Line [3] outputs the DataFrame to the terminal.
  • Line [4] creates a pivot table from the DataFrame based on the Population and Capital columns. The output saves to result.
  • Line [5] outputs the result to the terminal.

Output:

df

  Country Capital Population Area
0 Germany Berlin    83783942  357021
1 France   Paris    67081000  551695
2 Spain  Madrid    47431256  498511
3 Italy    Rome    60317116  301338
4 Poland  Warsaw    38383000  312685

result

Capital Berlin Madrid Paris Rome Warsaw
Population 83783942  47431256  67081000  60317116  38383000

For this example, a comma-delimited CSV file is read in. A pivot table is created based on selected parameters. Notice the max function.

Code – Example 2

df = pd.read_csv('countries.csv')
df = df.head(5)

result = pd.pivot_table(df, values='Population', columns='Capital', aggfunc=[max])
print(result)
  • Line [1] reads in a comma-separated CSV file and saves to a DataFrame (df).
  • Line [2] saves the first five (5) rows of the CSV file to df (over-writing df).
  • Line [3] creates a pivot table from the DataFrame based on the Population and Capital columns. The max population is a parameter of aggfunc. The output saves to result.
  • Line [4] outputs the result to the terminal.

Output:

result

  max        
Capital Berlin Madrid Paris Rome Warsaw
Population 83783942  47431256  67081000  60317116  38383000

DataFrame reorder_levels()

The reorder_levels() method re-arranges the index of a DataFrame/Series. This method can not contain any duplicate level(s) or drop level(s).

The syntax for this method is as follows:

DataFrame.reorder_levels(order, axis=0)
Parameter Description
order This parameter is a list containing the new order levels. These levels can be a position or a label.
axis If zero (0) or index is selected, apply to each column. Default is 0 (column). If zero (1) or columns, apply to each row.

For this example, there are five (5) students. Each student has some associated data with it. Grades generate by using np.random.randint().

index = [(1001, 'Micah Smith', 14), (1001, 'Philip Jones', 15), 
         	(1002, 'Ben Grimes', 16), (1002, 'Alicia Heath', 17), (1002, 'Arch Nelson', 18)]
m_index = pd.MultiIndex.from_tuples(index)
grades_lst = np.random.randint(45,100,size=5)
df = pd.DataFrame({"Grades": grades_lst}, index=m_index)
print(df)

result = df.reorder_levels([1,2,0])
print(result)
  • Line [1] creates a List of tuples. Each tuple contains three (3) values. The output saves to index.
  • Line [2] creates a MultiIndex from the List of Tuples created on line [1] and saves to m_index.
  • Line [3] generates five (5) random grades between the specified range and saves to grades_lst.
  • Line [4] creates a DataFrame from the variables on lines [1-3] and saves to df.
  • Line [5] outputs the DataFrame to the terminal.
  • Line [6] re-orders the levels as specified. The output saves to result.
  • Line [7] outputs the result to the terminal.

Output:

df

      Grades
1001 Micah Smith 14 52
  Philip Jones 15 65
1002 Ben Grimes 16 83
  Alicia Heath 17 99
  Arch Nelson  18 78

result

      Grades
Micah Smith 14 1001 52
Philip Jones 15 1001 65
Ben Grimes 16 1002 83
Alicia Heath 17 1002 99
Arch Nelson  18 1002 78

DataFrame sort_values()

The sort_values() method sorts (re-arranges) the elements of a DataFrame.

The syntax for this method is as follows:

DataFrame.sort_values(by, axis=0, ascending=True, inplace=False, kind='quicksort', na_position='last', ignore_index=False, key=None)
Parameter Description
by This parameter is a string or a list of strings. These comprise the index levels/columns to sort. Dependent on the selected axis.
axis If zero (0) or index is selected, apply to each column. Default is 0 (column). If zero (1) or columns, apply to each row.
ascending By default, True. Sort is conducted in ascending order. If False, descending order.
inplace If False, create a copy of the object. If True, the original object updates. By default, False.
kind Available options are quicksort, mergesort, heapsort, or stable. By default, quicksort. See numpy.sort for additional details.
na_position Available options are first and last (default). If the option is first, all NaN values move to the beginning, last to the end.
ignore_index If True, the axis numbering is 0, 1, 2, etc. By default, False.
key This parameter applies the function to the values before a sort. The data must be in a Series format and applies to each column.

For this example, a comma-delimited CSV file is read in. This DataFrame sorts on the Capital column in descending order.

df = pd.read_csv('countries.csv')
result = df.sort_values(by=['Capital'], ascending=False)
print(result)
  • Line [1] reads in a comma-delimited CSV file and saves to df.
  • Line [2] sorts the DataFrame on the Capital column in descending order. The output saves to result.
  • Line [3] outputs the result to the terminal.

Output:

  Country Capital Population Area
6 USA  Washington   328239523   9833520
4 Poland      Warsaw    38383000    312685
3 Italy        Rome    60317116    301338
1 France       Paris    67081000    551695
5 Russia      Moscow   146748590  17098246
2 Spain      Madrid    47431256    498511
8 India       Dheli  1352642280   3287263
0 Germany Berlin    83783942    357021
7 India Beijing  1400050000   9596961

DataFrame sort_index()

The sort_index() method sorts the DataFrame.

The syntax for this method is as follows:

DataFrame.sort_index(axis=0, level=None, ascending=True, inplace=False, kind='quicksort', na_position='last', sort_remaining=True, ignore_index=False, key=None)
Parameter Description
axis If zero (0) or index is selected, apply to each column. Default is 0 (column). If zero (1) or columns, apply to each row.
level This parameter is an integer, level name, or a list of integers/level name(s). If not empty, a sort is performed on values in the selected index level(s).
ascending By default, True. Sort is conducted in ascending order. If False, descending order.
inplace If False, create a copy of the object. If True, the original object updates. By default, False.
kind Available options are quicksort, mergesort, heapsort, or stable. By default, quicksort. See numpy.sort for additional details.
na_position Available options are first and last (default). If the option is first, all NaN values move to the beginning, last to the end.
ignore_index If True, the axis numbering is 0, 1, 2, etc. By default, False.
key This parameter applies the function to the values before a sort. The data must be in a Series format and applies to each column.

For this example, a comma-delimited CSV file is read into a DataFrame. This DataFrame sorts on the index Country column.

df = pd.read_csv('countries.csv')
df = df.set_index('Country')
result = df.sort_index()
print(result)
  • Line [1] reads in a comma-delimited CSV file and saves to df.
  • Line [2] sets the index of the DataFrame to Country. The output saves to df (over-writing original df).
  • Line [3] sorts the DataFrame (df) on the indexed column (Country) in ascending order (default). The output saves to result.
  • Line [4] outputs the result to the terminal.

Output:

  Country Population Area
China Beijing  1400050000   9596961
France Paris    67081000    551695
Germany Berlin    83783942    357021
India Dheli  1352642280   3287263
Italy Rome    60317116    301338
Poland Warsaw    38383000    312685
Russia Moscow   146748590  17098246
Spain Madrid    47431256    498511
USA Washington   328239523   9833520

Finxter