Dataframe (pandas)

09 Dec 2023 in Study / Computer science on Pandas

pandas

• One data structure of pandas
• Like matrix, there are rows (each row (= a record) name : index) and columns
• Creation
  

import pandas as pd

df1 = pd.DataFrame({"math" : [90, 74, 65], "biology" : [100, 60, 97], "chemistry" : [85, 100, 78]}, index = ["Samoh", "Rivera", "Cho"]) # if you don't insert index content, the default index is integer index

df2 = pd.DataFrame([[90, 100, 85], [74, 60, 100], [65, 97, 78]], index = ["Samoh", "Rivera", "Cho"], columns = ["math", "biology", "chemistry"])

• In df2, [90, 100, 85] is a record
  
• A person name : a single index
  
• A subject name : a single columns
  
• We can make dataframe like this way
  • df = pd.DataFrame([Series1, Series2, …], index = [“index0”, “index1”, …])

Change Index & Columns

• print(df.index) : check the index of df
• print(df.columns) : check the columns’ names
• change the names of all of things in index & columns
  • df.index = [i0, i1, i2, ….]
  • df.columns = [c0, c1, c2, ….]
• only change the parts that we wanna change
  • Use “rename”
    • origin source changes (X) → just return
    • If we wanna change the origin dataframe, we insert “inplace=True” as a parameter
  • df.rename(index={originI0 : newI0, originI1 : newI1, originI2 : newI2, …})
  • df.rename(columns={originC0 : newC0, originC1 : newC1, …})

Deletion of Rows & Columns

• Use “drop”
  • origin source changes (X) → just return
  • If we wanna change the origin dataframe, we insert “inplace=True” as a parameter
• df.drop(‘index0’, axis=0) (remove a row whose name is index0, default : axis=0)
• df.drop([‘index0’, ‘index3’], axis=0) (remove 2 rows)
• df.drop(‘column0’, axis=1) (remove a column whose name is column0)
• df.drop([‘column0’, ‘column3’], axis=1) (remove 2 columns)

Choose Rows

• Use “loc” (index label) & “iloc” (integer index)
• Choose a single row
  • df.loc[‘index0’] (return Series corresponding to index0)
  • df.loc[0] (return Series corresponding to first index)
• Choose multiple rows
  • Double times [] & Slicing returns dataframe
    • df.loc[‘index0’] != df.loc[[‘index0’]]
  • df.loc[ [‘index0’, ‘index3’] ] (return dataframe corresponding to index0 & index3)
  • df.loc[‘index0’ : ‘index3’] (return dataframe corresponding to index0 & index1 & index2 & index3)
  • df.iloc[[0, 3]] (return dataframe corresponding to first and fourth index)
  • df.iloc[0 : 4] (return dataframe corresponding to first, second, third, and fourth index)
    • df.iloc[0 : 4 : 2] (return dataframe corresponding to first, third index)
    • df.iloc[ : : -1] (return dataframe corresponding to last, last - 1, last - 2, …, first index)

Choose Columns

• Choose a single column
  • df[‘column1’] (return Series corresponding to column1)
  • df.column2 (return Series corresponding to column2)
• Choose multiple columns
  • df[ [‘column1’, ‘column2’, ‘column5’] ] (return dataframe corresponding to column1, column2, and column5)

Choose Element

• df.(i)loc[row information, column information]
  • Example
    • df.iloc[1, 2] (return a single element corresponding to second index and third column)
    • df.loc[‘index2’, ‘column2’] (return a single element corresponding to index2 and column2)
    • df.iloc[1, 2 : 5] (return Series corresponding to second index and third, fourth, fifth column)
    • df.loc[‘index2’, [‘column1’, ‘column3’]] (return Series corresponding to index2 and column1, column3)
    • df.iloc[1 : 3, 2 : 5] (return dataframe corresponding to second, third index and third, fourth, fifth column)
    • df.loc[‘index2’ : ‘index4’, [‘column1’, ‘column3’]] (return dataframe corresponding to index2, index3, index4 and column1, column3)
    • df.iloc[:, 2 : 5] (return dataframe corresponding to second index and third, fourth, fifth column)
    • df.loc[:, [‘column1’, ‘column3’]] (return dataframe corresponding to column1, column3)
• df.(i)loc[row information][column information]

Add Row

• New row has only one value as element
  • df.loc[“index5”] = 0 (add a row named as index5 and this row has only 0 as element)
• New row has various values as element
  • df.loc[“index5”] = [2, 4, 5, 7] (Series is also possible) (add a row named as index5 and [2, 4, 5, 7] is a record)

Add Column

• New column has only one value as element
  • df[“column5”] = 0 (add a column named as column5 and this column has only 0 as element)
• New row has various values as element
  • df[“column5”] = [2, 4, 5, 7] (Series is also possible) (add a column named as column5)

Change Element

• (Choose parts that we wanna change) = newValue
  • Example
    • df.iloc[1, 2] = 50 (Change the original value located on second index, thrid column to 50)
    • df.loc[‘index1’, [‘column2’, ‘column4’]] = [40, 60] (Change the original values located on (second index, column2), (second index, column4) to 40 and 60)

Transpose

• row → column, column → row
• df.transpose()
• df.T

Head & Tail

• df.head(number) (return dataframe with n rows starting from the first row, default : number == 5)
• df.tail(number) (return dataframe with n rows starting from the last row, default : number == 5)