Pandas

Pandas is a high-level data manipulation tool developed by Wes McKinney.

It is built on the top of Numpy package

Pandas key data structure is called the Series and DataFrame.

DataFrames allow you to store and manipulate tabular data in rows of observations and
columns of variables.

Pandas is an open source Python package that is most widely used for data
science/data analysis and machine learning tasks

What Can You Do With DataFrames Using Pandas?

Series : series are similar to numpy array except we can give named or
datetime index instead of numerical index

import numpy as np

import pandas as pd

lable =['a','b','c']

lst =[10,20,30]

arr =np.array([10,20,30])

dis ={'a':10,'b':20,'c':30}

pd.Series(lst)
pd.Series(lst,lable)

pd.Series(arr,lable)

pd.Series(dis)

pd.Series([sum,print,len])

ser1 =pd.Series([1,2,3,4],['USA','CHAINA','FRANCE','GERMANY'])

ser2 =pd.Series([1,2,3,4],['USA','CHAINA','INDIA','SINGAPOOR'])

ser1

ser2

ser1['USA']

ser1 + ser2

Data frames which is but directly top of series which is used in financial data

The numpy.random.randn() function creates an array of specified shape and fills it with random

import numpy as np

import pandas as pd

from numpy.random import randn

np.random.seed(101)

df =pd.DataFrame(randn(5,4),['A','B','C','D','E'],['W','X','Y','Z'])

columns are series all sharing common index

df['W']

type(df['W'])

type(df)

df.w

df[['W','X']]

adding new column to data frame

df['new'] =df['Y']+df['Z']
df.drop('new')

df.drop('new',axis=1)

df

df.drop('new',axis=1,inplace=True)

df.drop('E',inplace=True)

selecting row in two ways

df.loc['A']

df.iloc[2]

df.loc[['A','B']]

subset of rows and columns

select row a, b and column w,y

df.loc[['A','B'],['W','Y']]

df.iloc[2:,:]

df.iloc[2:,2:]

df.iloc[2:,:2]

df.iloc[:2,:2]

f.iloc[1:3,1:3]

df.iloc[-2:,-2:]

df.iloc[0:2,0:2]

df >0

booldf =df >0

df[booldf]

df[df>0]

df['W']>0

df[df['W']>0]
resultdf =df[df['W']<0]

resultdf

resultdf[['X','Z']]

Instead of doing two steps

df[df['W']<0][['X','Z']]

df[(df['W']<0) & (df['Y'] >0)]

df[(df['W']<0) | (df['Y'] >0)]

df.reset_index()

lst=['TN','AP','KA','MH','TS']

df['STATE']=lst

Multi level data frame

outside =['G1','G1','G1','G2','G2','G2']

inside =[1,2,3,1,2,3]

hier_index=list(zip(outside,inside))

hier_index=pd.MultiIndex.from_tuples(hier_index)

df =pd.DataFrame(randn(6,2),hier_index,['A','B'])

df.loc['G1']

df.loc['G1']['A']

df.index.names

df.index.names=['Groups','Num']

df.loc['G2'].loc[2]['B']

Cross Section

df.xs('G1')

df.xs(1,level='Num')

df.xs(('G1',2))
Missing data

d ={'A':[1,2,np.nan],'B':[5,np.nan,np.nan],'C':[1,2,3]}

df = pd.DataFrame(d)

df.dropna()

df.dropna(axis=1)

df.dropna(thresh=2)

fill value

df.fillna(value=0)

df['A'].fillna(df['A'].mean())

df['A'].fillna(df['A'].mean(),inplace=True)

Grouping

d ={'Company':['GOOG','GOOG','MSFT','MSFT','FB','FB'],

'Person':['RAM','SHAM','SUNIL','SUDEEP','RAHEEM','SHEETAL'],

'Sales':[250,400,200,150,350,100]}

df =pd.DataFrame(d)

bycomp.mean()

bycomp.max()

bycomp.std()

bycomp.min()

bycomp.sum()

bycomp.sum().loc['FB']

bycomp.describe()

bycomp.describe().transpose()

df.groupby('Company').describe().transpose()['FB']

Merging , joining,Concatination
df1 =pd.DataFrame({'A':['A0','A1','A2','A3'],

'B':['B0','B1','B2','B3'],

'C':['C0','C1','C2','C3']},

index =[0,1,2,3])

df2=pd.DataFrame({'A':['A4','A5','A6','A7'],

'B':['B4','B5','B6','B7'],

'C':['C4','C5','C6','C7']},

index =[4,5,6,7])

df3=pd.DataFrame({'A':['A8','A9','A10','A11'],

'B':['B8','B9','B10','B11'],

'C':['C8','C9','C10','C11']},

index =[8,9,10,11])

concatinate

pd.concat([df1,df2,df2])

pd.concat([df1,df2,df2],axis=1)

left =pd.DataFrame({'key':['K0','K1','K2','K3'],

'A':['A0','A1','A2','A3'],

'B':['B0','B1','B2','B3']})

right =pd.DataFrame({'key':['K0','K1','K2','K3'],

'C':['C0','C1','C2','C3'],

'D':['D0','D1','D2','D3']})

pd.merge(left,right,how='inner',on='key')

emp=pd.DataFrame({'EMPNO':['E001','E0002','E003','E004'],

'ENAME':['BABJEE','RAM','SUNIL','SHAM'],

'DEPTNO':[10,10,20,30]})
dept=pd.DataFrame({'Dname':['Accounts','Admin','It'],'DEPTNO':[10,20,50]})

pd.merge(emp,dept,how='inner',on='DEPTNO')

emp=pd.DataFrame({'EMPNO':['E001','E0002','E003','E004'],

'ENAME':['BABJEE','RAM','SUNIL','SHAM']},

index =[10,10,20,30])

dept=pd.DataFrame({'DNAME':['Accounts','Admin','It'],

'LOCATION':['CHENNAI','MUMBAI','PUNE'] },

index=[10,20,50])

emp.join(dept,how='inner')

emp.join(dept,how='outer')

df =pd.DataFrame({'Col1':[1,2,3,4],

'Col2':[444,555,666,444],

'Col3':['abc','def','ghi','xyz']})

df.head(2)

df.tail(2)

df['Col2'].unique()

len(df['Col2'].unique())

df['Col2'].nunique()

df['Col2'].value_counts()

df[df['Col1']>2]

df[(df['Col1']>2) & (df['Col2']==444)]

df['Col1'].sum()

Customs function
def times2(x):

retrun x*x
df['Col1'].apply(times2)

calling built-in functions

df['Col3'].apply(len)

df['Col2'].apply(lambda x: x *x)

df.drop('Col1',axis=1)

df.columns

df.index

df.sort_values(by='Col2',ascending=False)

df.isnull()

input and output

pwd

pd.read_csv('d:\\demo\example.csv')

pd.read_excel("d:\demo\example.xlsx")

df.to_csv("d://demo/myoutput.csv",index=False)

pd.read_excel("d:\\demo\\example.xlsx",sheet_name='Sheet1')

df.to_excel("d:\\demo\\example1.xlsx",sheet_name='Sheet2',index=False)

table_MN = pd.read_html('https://en.wikipedia.org/wiki/Minnesota',match='Election results from

import pandas as pd

from sqlalchemy import create_engine

cnx = create_engine('mysql+pymysql://root:admin123@localhost:3306/demo').connect()

sql = 'select * from customers'

df = pd.read_sql(sql, cnx)
The Pandas datareader is a sub package that allows one to create a dataframe from
various internet datasources, currently including:

 Yahoo! Finance
 Google Finance
 St.Louis FED (FRED)
 Kenneth French’s data library
 World Bank
 Google Analytics

pip install pandas-datareader

import pandas_datareader.data as web

import datetime as dt

start=dt.datetime(2015,1,1)

end=dt.datetime(2015,12,31)

facebook =web.DataReader('FB','yahoo',start,end

Pandas time series

majority of data in financial analysis is time series

datatime index

import pandas as pd

import numpy as np

from datetime import datetime

first_two =[datetime(2017,1,1),datetime(2017,1,2)]

dt_ind =pd.DatetimeIndex(first_two)

data =np.random.randn(2,2)

df =pd.DataFrame(data,dt_ind,['a','b'])

df.index.argmax()

df.index.argmin()

df.index.max()
Time resampling

df = pd.read_csv("d://demo//walmart_stock.csv")

df.head()

df.info()

df['Date']=pd.to_datetime(df['Date'])

df.info()

df.set_index('Date',inplace=True)

df.index()