[Python] How to get & change rows / columns / values from a table.
[Python] How to get & change rows / columns / values from a table
From a table created with Pandas DataFrame ・ ** How to extract the specified element ** -** Summary of how to change the value of the specified element **.
**table of contents**
- [Extract column](# 1 extract column)
- [Extract column name list](# Extract column name list)
- [Extract by column name](# Extract by column name)
- [Extract by column number](# Extract by column number)
- [Extract Rows](# 2 Extract Rows)
- [Extract line name list](# Extract line name list)
- [Extract by line name](# Extract by line name)
- [Extract by line number](# Extract by line number)
- [Extract Element](# 3 Extract Element)
- [Change data (change by specifying the element)](# 4 Change data)
## Table to use Extract by specifying the elements from the table created with 4 rows and 4 columns.
▼ Table to be used (assigned to variable "df")
Table creation code
import pandas as pd
listA = ['AAA', 'BBB', 'CCC', 'DDD']
listB = ['EEE', 'FFF', 'GGG', 'HHH']
listC = ['III', 'JJJ', 'KKK', 'LLL']
listD = ['MMM', 'NNN', 'OOO', 'PPP']
df = pd.DataFrame(listA)
df[1] = listB
df[2] = listC
df[3] = listD
df.columns = ['aaa', 'bbb', 'ccc', 'ddd']
df.index = [111,222,333,444]
df
For details on how to create a table, click here ()
## 1. 1. Column extraction ① Extraction of column name list ② Extract by column name ③ Extract by column number
▼ ① Extract the column name list
Use the .columns method
└ Extraction of column name list
└ Output is Index / Int64Index type, etc.
└ Column type (dtype) is also output
[Table to use and its code](#Table to use)
Extract column names
df.columns
#output
# Index(['aaa', 'bbb', 'ccc', 'ddd'], dtype='object')
The output will be of type Index.
▼ Index type can be list.
Output as list type
list(df.columns)
#output
# ['aaa', 'bbb', 'ccc', 'ddd']
** ▼ Output only the heading of the specified column number ** `df.columns[n]` └ "n": Sequence number
Extract by column number
df.columns[2]
#output
# 'ccc'
### ② Extract by column name There are three ways to write by column name ②-1. df ['column name'] ②-2. Df. Column name ②-3. df.loc [:, ['column name']]
- Df is a variable with a table
- No quotation marks or parentheses are required for the column name of "df. Column name". An error will occur if the column name is not described as it is.
Extract by column name ①
df['bbb']
#output
111 EEE
222 FFF
333 GGG
444 HHH
Name: bbb, dtype: object
Extract by column name ②
df.bbb
#output
111 EEE
222 FFF
333 GGG
444 HHH
Name: bbb, dtype: object
Quotation marks are not required even for character strings. If you turn it on, an error will occur.
error
df.'bbb'
#output
# invalid syntax
** ▼ Extract with loc method **
df.loc[:, ['aaa', 'bbb',,,]]
└ "loc []": loc method
└ ":,": No line name specified
└ "['aaa','bbb' ,,,]": Specify the column name you want to extract.
Extract by column name ③ (loc)
df.loc[:, ['bbb', 'ccc']]
bbb ccc
111 EEE III
222 FFF JJJ
333 GGG KKK
444 HHH LLL
- If there is only one column, [] may or may not be present.
- Output format changes (column name, dtype, etc.)
df.loc[:, 'aaa']=df.loc[:, ['aaa']]
### ③ Extract by column number `iloc[:,[n,x,,,]]` └ "iloc []": Specify by column number └ ":,": No line specified └ "n" "x": column number └ Slice [n: x] cannot be used (error)
Extract with iloc
df.iloc[:, [1,2]]
#output
bbb ccc
111 EEE III
222 FFF JJJ
333 GGG KKK
444 HHH LLL
- If there is only one column, [] may or may not be present.
- Output format changes (column name, dtype, etc.)
df.loc[:, 2]=df.loc[:, [2]]
** ▼ Extract only one column **
Extract with iloc (only one column ①)
df.iloc[:, 2]
#output
111 III
222 JJJ
333 KKK
444 LLL
Name: ccc, dtype: object
Extract with iloc (only one column ②)
df.iloc[:, [2]]
#output
ccc
111 III
222 JJJ
333 KKK
444 LLL
## 2. Extraction of rows ① Extraction of line name list ② Extract by line name ③ Extract by line number
▼ ① Extraction of line name list
Use the .index method
└ Extraction of column name list
└ Output is Index / Int64Index type, etc.
└ Column type (dtype) is also output
[Table to use and its code](#Table to use)
Extract line name
df.index
#output
# Int64Index([111, 222, 333, 444], dtype='int64')
In this case, the output will be Int64Index type.
▼ index / int64 Index type can be list.
Output as list type
list(df.index)
#output
# [111, 222, 333, 444]
** ▼ Output only the heading of the specified line number ** `df.index[n]` └ "n": Line number
Extract by line number
df.index[2]
#output
# '333'
### ② Extract by line name `df.loc[['aaa', 'bbb',,,,]]` └ "df": Variable with table └ ".loc []": How to specify row / column names └ "aaa" "bbb": line name
Extract by line name
df.loc[[111,333,444]]
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
333 CCC GGG KKK OOO
444 DDD HHH LLL PPP
** ▼ Extract after the specified line ** `df.loc [line name:]`
Extract after the specified line
df.loc[333:]
#output
aaa bbb ccc ddd
333 CCC GGG KKK OOO
444 DDD HHH LLL PPP
** ▼ In case of only one line ** -One line is described in two patterns. -The output format changes (column name, dtype, etc.) `df.loc[[111]]` `df.loc[111]` └ Double parentheses are easier to see.
** ▼ Example of `df.loc [[111]]` **
Extract with loc (only one column ①)
df.loc[[111]]
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
** ▼ Example of df.loc [111] **
Extract with loc (only one column ②)
df.loc[111]
#output
aaa AAA
bbb EEE
ccc III
ddd MMM
Name: 111, dtype: object
③ Extract by line number
③-1. df[n:m:l]
③-2. df.iloc[[n,m,,,,]]
（3-1）df[n:m:l]
Specify the rows to extract using slices.
└ "df": Variable with table
└ "n": Start value
└ "m": End price
└ "l": amount of change
・ "N" and "l" can be omitted.
** ▼ Example: df [1: 3] ** Extract the 1st and 2nd lines. The 3rd line is not included (because the processing ends at the 3rd line)
Extraction of rows
df[1:3]
#output
aaa bbb ccc ddd
222 BBB FFF JJJ NNN
333 CCC GGG KKK OOO
** ▼ Example: Specify the amount of change and extract **
df[1:5:2]
From the 1st to 4th lines, extract the lines with numbers incremented by 2.
Extraction of rows
df[1:5:2]
#output
aaa bbb ccc ddd
222 BBB FFF JJJ NNN
444 DDD HHH LLL PPP
** ▼ When the initial value is omitted ** The initial value is regarded as 0, and the lines up to the specified number (one before) from the top can be extracted.
df[:3]
Extract from 0 to 2nd line.
Extraction of rows
df[:3]
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
222 BBB FFF JJJ NNN
333 CCC GGG KKK OOO
### (3-2) Extract with iloc
df.iloc[[n,m,,,]]
└ "df": Variable with table
└ ".iloc []": Method specified by row / column number
└ "n" "m": Specify line number
- [] Enclosing the line number is required
- Slices cannot be used.
▼ Example: `df.iloc [[1,3]]`
Specified by line number
df.iloc[[1,3]]
#output
aaa bbb ccc ddd
222 BBB FFF JJJ NNN
444 DDD HHH LLL PPP
#### When extracting only one line There are two patterns of description. `df.iloc[[n]]` `df.iloc[n]`
Difference in the number of []. The output result is different for each. └ The output is easier to see with df.iloc [[n]].
** ▼ Example: df.iloc [[n]] **
Output for double parentheses.
Specified by line number
df.iloc[[3]]
#output
aaa bbb ccc ddd
444 DDD HHH LLL PPP
** ▼ Example: `df.iloc [n]` ** Output for single parentheses.
Specified by line number
df.iloc[3]
#output
aaa DDD
bbb HHH
ccc LLL
ddd PPP
Name: 444, dtype: object
## 3. 3. Extraction of elements There are four main methods to use. ①loc ②iloc ③at ④iat
** ■ Can be broadly classified into "loc" and "at" ** -"Loc": Specify by range. You can also get multiple rows and columns. -"At": Get only one element pinpoint.
〇Supplement: "at" is faster.
** ■ Presence or absence of "i" ** With or without the "i" at the beginning, you can choose to specify by name or by number.
・ None: Matrix name ・ Yes: Matrix number
### ▼ Extraction source table 
Table creation code
import pandas as pd
listA = ['AAA', 'BBB', 'CCC', 'DDD']
listB = ['EEE', 'FFF', 'GGG', 'HHH']
listC = ['III', 'JJJ', 'KKK', 'LLL']
listD = ['MMM', 'NNN', 'OOO', 'PPP']
df = pd.DataFrame(listA)
df[1] = listB
df[2] = listC
df[3] = listD
df.columns = ['aaa', 'bbb', 'ccc', 'ddd']
df.index = [111,222,333,444]
df
(3-1) loc
df.loc ['row name','column name']
loc
df.loc[333,'bbb']
#output
# 'GGG'
(3-2) iloc `df.iloc ['row number','column number']`
iloc
df.iloc[2,1]
#output
# 'GGG'
(3-3) at `df.at ['row name','column name']`
at
df.at[333,'bbb']
#output
# 'GGG'
(3-4) iat `df.at ['row number','column number']`
iat
df.iat[2,1]
#output
# 'GGG'
## 4. Data change ### Change by specifying an element If you specify an element and substitute it, the data will be replaced. There are four main methods to use.
①loc ②iloc ③at ④iat
Details
** ■ Can be broadly classified into "loc" and "at" **
-"Loc": Specify by range. You can also get multiple rows and columns.
-"At": Get only one element pinpoint.
** ■ Presence or absence of "i" **
With or without the "i" at the beginning, you can choose to specify by name or by number.
・ None: Matrix name
・ Yes: Matrix number
** ■ Presence or absence of "i" ** With or without the "i" at the beginning, you can choose to specify by name or by number.
・ None: Matrix name ・ Yes: Matrix number
### Original table Use the table below to rewrite the data. 
Table creation code
import pandas as pd
listA = ['AAA', 'BBB', 'CCC', 'DDD']
listB = ['EEE', 'FFF', 'GGG', 'HHH']
listC = ['III', 'JJJ', 'KKK', 'LLL']
listD = ['MMM', 'NNN', 'OOO', 'PPP']
df = pd.DataFrame(listA)
df[1] = listB
df[2] = listC
df[3] = listD
df.columns = ['aaa', 'bbb', 'ccc', 'ddd']
df.index = [111,222,333,444]
df
(4-1) loc
loc
df.loc[333,'bbb'] = '@@@'
df
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
222 BBB FFF JJJ NNN
333 CCC @@@ KKK OOO
444 DDD HHH LLL PPP
(4-2) iloc
iloc
df.iloc[2,1] = '★★'
df
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
222 BBB FFF JJJ NNN
333 CCC ★★ KKK OOO
444 DDD HHH LLL PPP
(4-3) at
at
df.at[333,'bbb'] = 999
df
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
222 BBB FFF JJJ NNN
333 CCC 999 KKK OOO
444 DDD HHH LLL PPP
(4-4) iat
iat
df.iat[2,1] = '※※'
df
#output
aaa bbb ccc ddd
111 AAA EEE III MMM
222 BBB FFF JJJ NNN
333 CCC ※※ KKK OOO
444 DDD HHH LLL PPP
### ■ Change row name and column name ① When creating a table Use DataFrame options · Columns: columns option · Row: index option
② Later (collectively) · Columns: columns method -Line: index method
③ Later (designated) -Rename method └ "columns = {}": Change column name └ "index = {}": Change line name
** ▶ How to change the row name and column name here **
> Official page -[Pandas.DataFrame.loc](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.loc.html#pandas.DataFrame.loc) -[Pandas.DataFrame.iloc](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.iloc.html#pandas.DataFrame.iloc) -[Pandas.DataFrame.at](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.at.html#pandas.DataFrame.at) -[Pandas.DataFrame.iat](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.iat.html#pandas.DataFrame.iat)
[Back to top](# How to get and change matrix values from python table)