Basic data structure

1. Series

For a Series, the most commonly used attributes are values, index, name, and dtype

s = pd.Series(np.random.randn(5),index=['a','b','c','d','e'],name='This is a Series',dtype='float64')
s

a   -0.152799
b   -1.208334
c    0.668842
d    1.547519
e    0.309276
Name: This is a Series, dtype: float64

Accessing Series properties

s.values
array([-0.15279875, -1.20833379,  0.6688421 ,  1.54751933,  0.30927643])
s.name
'This is a Series'
s.index
Index(['a', 'b', 'c', 'd', 'e'], dtype='object')
s.dtype
dtype('float64')

DataFrame

df = pd.DataFrame({'col1':list('abcde'),'col2':range(5,10),'col3':[1.3,2.5,3.6,4.6,5.8]},
                 index=list('one two three four five'))
df

Take a column from the DataFrame as Series

df['coll']

Modify row or column names

df.rename(index={'one':'one'},columns={'col1':'new_col1'})

Calling properties and methods

df.index
Index(['one', 'two', 'three', 'four', 'five'], dtype='object')
df.columns
Index(['col1', 'col2', 'col3'], dtype='object')
df.values
array([['a', 5, 1.3],
       ['b', 6, 2.5],
       ['c', 7, 3.6],
       ['d', 8, 4.6],
       ['e', 9, 5.8]], dtype=object)
df.shape
(5, 3)

Alignment properties of indexes

This is a very powerful feature in Pandas, for example

df1 = pd.DataFrame({'A':[1,2,3]},index=[1,2,3])
df2 = pd.DataFrame({'A':[1,2,3]},index=[3,1,2])
df1-df2

Deletion and addition of columns

For deletion, you can use the drop function or del or pop
Of course, it should be noted that when using the drop function, if the parameter inplace=True, it will be directly changed in the original DataFrame

df.drop(index='five',columns='col1')

df['col1']=[1,2,3,4,5]
del df['col1']
df

The pop method operates directly on the original DataFrame and returns the deleted column, which is similar to the pop function in python

df['col1']=[1,2,3,4,5]
df.pop('col1')

You can add new columns directly or use the assign method
However, the assign method will not modify the original DataFrame

df1['B']=list('abc')

Select columns by type

df.select_dtypes(include=['number']).head()
df.select_dtypes(include=['float']).head()

Finally, the T symbol can be transposed

2. Common functions

head and tail

Return to the original line: Head()
Return to the last line: Tail()
If no parameter is added in the default bracket, the first five lines or the last five lines will be selected by default

unique and nunique

nunique shows how many unique values there are
Unique displays all unique values

count and value_counts

count returns the number of elements in a row or column
value_counts returns the number of each element

info and describe

The info function returns which columns are there, how many non missing values are there, and the type of each column
describe the statistics of numerical data by default

df.info()

df.describe()

You can also choose the quantile by yourself

df.describe(percentiles=[.05, .25, .75, .95])

For non numeric types, you can also use the describe function

idxmax and nlargest

The idxmax function returns the index of the maximum value, which is especially applicable in some cases. The idxmin function is similar
The nlargest function returns the first few large element values, and the function of nsmallest is similar

df['ciol1'].nlargest(3)

clip and replace

clip and replace are two types of replacement functions
clip is to truncate the number that exceeds or falls below some value
Replace is to replace some values

7. apply function

This function has a high degree of freedom, which I will explain specifically.