快速介绍Python数据分析库pandas的基础知识和代码示例

 import pandas as pd

 # Load data
 df = pd.read_csv('filename.csv') # From a CSV file
 df = pd.read_excel('filename.xlsx') # From an Excel file

 df.to_csv('filename.csv') # Write to a CSV file
 df.to_excel('filename.xlsx') # Write to an Excel file

 # Build data frame from inputted data
 df = pd.DataFrame(data = {'Name': ['Bob', 'Sally', 'Scott', 'Katie'],
  'Physics': [68, 74, 77, 78],
  'Chemistry': [84, 100, 73, 90],
  'Algebra': [78, 88, 82, 87]})

 #  Create a series from an iterable my_list
 my_list = [['Bob',78],
           ['Sally',91],
           ['Scott',62],
           ['Katie',78],
           ['John',100]]
 df1 = pd.Series(my_list) # Create a series from an iterable my_list

 df.head(3) # First 3 rows of the DataFrame

 df.tail(3) # Last 3 rows of the DataFrame

 #Append row to the dataframe, missing data (np.nan)
 new_row = {'Name':'Max', 'Physics':67, 'Chemistry':92, 'Algebra':np.nan}
 df = df.append(new_row, ignore_index=True)

 # List of series  
 list_of_series = [pd.Series(['Liz', 83, 77, np.nan], index=df.columns),
                 pd.Series(['Sam', np.nan, 94,70], index=df.columns ),
                 pd.Series(['Mike', 79,87,90], index=df.columns),
                 pd.Series(['Scott', np.nan,87,np.nan], index=df.columns),]
 # Pass a list of series to the append() to add multiple rows
 df = df.append(list_of_series , ignore_index=True)

 # Adding a new column to existing DataFrame in Pandas
 sex = ['Male','Female','Male','Female','Male','Female','Female','Male','Male']
 df['Sex'] = sex

 df.info() #Index, Datatype and Memory information

 # Check data type in pandas dataframe
 df['Chemistry'].dtypes
 >>> dtype('int64')#  Convert Integers to Floats in Pandas DataFrame
 df['Chemistry'] = df['Chemistry'].astype(float)
 df['Chemistry'].dtypes
 >>> dtype('float64')# Number of rows and columns
 df.shape
 >>> (9, 5)

 # View unique values and counts of Physics column
 df['Physics'].value_counts(dropna=False)

 df['Chemistry'] # Returns column with label 'Chemistry' as Series

 df[['Name','Algebra']] # Returns columns as a new DataFrame

 df.iloc[0] # Selection by position

df.iloc[:,1] # Second column 'Name' of data frame

df.iloc[0,1] # First element of Second column
>>> 68.0

# Rename columns
df = df.rename({'Name':'Student','Algebra':'Math'}, axis='columns')

# Checks for null Values, Returns Boolean Arrray
check_for_nan = df.isnull()

value = df.notnull() # Opposite of df2.isnull()

drop_null_row = df.dropna() # Drop all rows that contain null values

# Drop all columns that contain null values
drop_null_col = df.dropna(axis=1)

replace_null = df.fillna(0) # Replace all null values with 0

# Replace all null values with the mean (mean can be replaced with almost any function from the statistics module)
df = round(df.fillna(df.mean()),2)

one = df.replace(100,'A') # Replace all values equal to 1 with 'one'

fil_80 = df[df['Physics'] > 80]

fil = df[(df['Chemistry'] > 80) & (df['Math'] < 90)]

ascending = df.sort_values('Student')

descending = df.sort_values('Chemistry',ascending=False)

df.sort_values(['Physics','Chemistry'],ascending=[True,False])

group_by = df.groupby(['Sex']) # Returns a groupby object for values from one column
group_by.first() # Print the first value in each group

average = df.groupby(‘Sex’).agg(np.mean)

pivot_table = df.pivot_table(index='Sex',
                             values=['Physics','Chemistry'],
                             aggfunc=[len, np.mean, np.std])

df.describe() # Summary statistics for numerical columns

# Get a series containing maximum value of each row
max_row = df.max(axis=1)

# Get a series containing maximum value of each column without skipping NaN
max_col = df.max(skipna=False)