Day 1 16/12/19, 17:40

In [134]: import pandas as pd

import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline

In [135]: # LISTS/ARRAYS ARE FUNDAMENTAL DATATYPES

# LISTS/ARRAY START WITH SQUARE BRACKETS
# LISTS ARE MUTABLE OR APPENDABLE
numlist=[2,3,5,666,223,89.5]
print(numlist)
charlist=['aaa',"AAA",'c','john']
print(charlist)
booleanlist=[True,False,False,True]
print(booleanlist)

[2, 3, 5, 666, 223, 89.5]

['aaa', 'AAA', 'c', 'john']
[True, False, False, True]

In [136]: # TUPPLES ARE IMMUTABLE OBJECTS

# TUPPLES HAVE PARANTHESIS OR ROUND BRACKETS
numtupple=(23,34,24,25)
print(numtupple)
chartupple=('aaa',"BBBB",'CCC')
print(chartupple)

(23, 34, 24, 25)

('aaa', 'BBBB', 'CCC')

In [137]: # STRINGS are character based/text based

string="hello good morning"
print(string)

hello good morning

In [138]: # Sets created by placing items in curly brackets

# multiple datatypes int,float,string,tupple
# Sets are mutable
myset={1,"Hello",(1,2,3)}
print(myset)

# Ordered Output

{1, 'Hello', (1, 2, 3)}

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 1 of 19
Day 1 16/12/19, 17:40

In [139]: # DICTIONARIES - INTERGER OR STRINGS WITH KEYS

# WITH CURLY BRACKETS
dict1={'Name':'John','Age':24,'Gender':'M'}
print(dict1)
dict1['Name']='Pete'
print(dict1)

{'Name': 'John', 'Age': 24, 'Gender': 'M'}

{'Name': 'Pete', 'Age': 24, 'Gender': 'M'}

In [140]: mydict2={}
mydict2['Key1']=[1,2]
mydict2['key2']=['G','F']
print(mydict2)

{'Key1': [1, 2], 'key2': ['G', 'F']}

In [141]: # Indexing is done using square brackets

# PYTHON STARTS WITH 0
print(numlist[0]) # Positive Indexing
print(numlist[-5]) # Negative Indexing
print(dict1.get('Name')) # DICTIONARIES 'get'

print(dict1['Name']) # DICTIONARIES SQUAREBRACKET

print(numtupple[3])

2
3
Pete
Pete
25

In [142]: # FOR IMPORTING DATA FILEPATH WITH FILENAME &

# EXTENSION & CHANGE SLASHES TO OPPOSITE SIDE
# in DOUBLE QUOTES
bankchurn=pd.read_csv("/Users/rajeshprabhakar/Downloads/Churn_Model
ling.csv")

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 2 of 19
Day 1 16/12/19, 17:40

In [143]: bankchurn.head() # FIRST 5 ROWS of DATA DEFAULT

Out[143]:
RowNumber CustomerId Surname CreditScore Geography Gender Age Tenure Balance

0 1 15634602 Hargrave 619 France Female 42 2 0.00

1 2 15647311 Hill 608 Spain Female 41 1 83807.86

2 3 15619304 Onio 502 France Female 42 8 159660.80

3 4 15701354 Boni 699 France Female 39 1 0.00

4 5 15737888 Mitchell 850 Spain Female 43 2 125510.82

In [144]: bankchurn.tail()

Out[144]:
RowNumber CustomerId Surname CreditScore Geography Gender Age Tenure Balan

9995 9996 15606229 Obijiaku 771 France Male 39 5

9996 9997 15569892 Johnstone 516 France Male 35 10 57369.

9997 9998 15584532 Liu 709 France Female 36 7

9998 9999 15682355 Sabbatini 772 Germany Male 42 3 75075.

9999 10000 15628319 Walker 792 France Female 28 4 130142.

In [145]: bankchurn.columns

Out[145]: Index(['RowNumber', 'CustomerId', 'Surname', 'CreditScore', 'Geogr

aphy',
'Gender', 'Age', 'Tenure', 'Balance', 'NumOfProducts', 'Has
CrCard',
'IsActiveMember', 'EstimatedSalary', 'Exited'],
dtype='object')

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 3 of 19
Day 1 16/12/19, 17:40

In [146]: bankchurn.dtypes
# DATA TYPES
# NUMBER - int64 & float64
# CHARACTER - object
# Date - Datetime
# Boolean - True False

Out[146]: RowNumber int64

CustomerId int64
Surname object
CreditScore int64
Geography object
Gender object
Age int64
Tenure int64
Balance float64
NumOfProducts int64
HasCrCard int64
IsActiveMember int64
EstimatedSalary float64
Exited int64
dtype: object

In [147]: bankchurn.shape # NUM of ROWS/OBSERVATIONS

# NUM of COLUMNS/VARIABLES

Out[147]: (10000, 14)

In [148]: bankchurn.describe().transpose()
# describe() - Basic DESCRIPTIVE STATISTICS

Out[148]:
count mean std min 25% 50%

RowNumber 10000.0 5.000500e+03 2886.895680 1.00 2500.75 5.000500e+03

CustomerId 10000.0 1.569094e+07 71936.186123 15565701.00 15628528.25 1.569074e+07

CreditScore 10000.0 6.505288e+02 96.653299 350.00 584.00 6.520000e+02

Age 10000.0 3.892180e+01 10.487806 18.00 32.00 3.700000e+01

Tenure 10000.0 5.012800e+00 2.892174 0.00 3.00 5.000000e+00

Balance 10000.0 7.648589e+04 62397.405202 0.00 0.00 9.719854e+04

NumOfProducts 10000.0 1.530200e+00 0.581654 1.00 1.00 1.000000e+00

HasCrCard 10000.0 7.055000e-01 0.455840 0.00 0.00 1.000000e+00

IsActiveMember 10000.0 5.151000e-01 0.499797 0.00 0.00 1.000000e+00

EstimatedSalary 10000.0 1.000902e+05 57510.492818 11.58 51002.11 1.001939e+05

Exited 10000.0 2.037000e-01 0.402769 0.00 0.00 0.000000e+00

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 4 of 19
Day 1 16/12/19, 17:40

In [149]: # EXPLORATORY DATA ANALYSIS (EDA) - BASIC

# UNDERSTANDING OF DATA
# DESCRIPTIVE STATISTICS+DATA VIZUALIZATION -EDA
# DESCRIPTIVE STATISTICS
# COUNT, MINIMUM, MAXIMUM
# MEASURES OF CENTRAL TENDENCY - MEAN, MEDIAN,
# MODE
# MEASURES OF DISPERSION - VARIANCE FROM MEAN
# RANGE, VARIANCE, STANDARD DEVIATION, QUARTILES
# PERCENTILES,DECILES
# MEASURES OF ASYMMETRY - SKEWNESS & KURTOSIS
# MEASURES OF RELATIONSHIP - COVARIANCE &
# CORRELATION

In [150]: # INDEXING OF ROWS AND COLUMNS - DATAFRAME

# CHARACTER INDEXING - BY COLUMN NAME
bankchurn.Balance.describe()

Out[150]: count 10000.000000

mean 76485.889288
std 62397.405202
min 0.000000
25% 0.000000
50% 97198.540000
75% 127644.240000
max 250898.090000
Name: Balance, dtype: float64

In [151]: bankchurn['Balance'].describe()

Out[151]: count 10000.000000

mean 76485.889288
std 62397.405202
min 0.000000
25% 0.000000
50% 97198.540000
75% 127644.240000
max 250898.090000
Name: Balance, dtype: float64

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 5 of 19
Day 1 16/12/19, 17:40

In [152]: bankchurn[['Balance','Tenure','Age']].describe()
# multi column indexing - double square brackets

Out[152]:
Balance Tenure Age

count 10000.000000 10000.000000 10000.000000

mean 76485.889288 5.012800 38.921800

std 62397.405202 2.892174 10.487806

min 0.000000 0.000000 18.000000

25% 0.000000 3.000000 32.000000

50% 97198.540000 5.000000 37.000000

75% 127644.240000 7.000000 44.000000

max 250898.090000 10.000000 92.000000

In [153]: # NUMERICAL INDEXING - COLUMN NUMBER

# PYTHON STARTS WITH 0
bankchurn.iloc[:,[6,7,8]].describe()
# :, before represents COLUMN Indexing
# IF :, not given default is ROW INDEXING

Out[153]:
Age Tenure Balance

count 10000.000000 10000.000000 10000.000000

mean 38.921800 5.012800 76485.889288

std 10.487806 2.892174 62397.405202

min 18.000000 0.000000 0.000000

25% 32.000000 3.000000 0.000000

50% 37.000000 5.000000 97198.540000

75% 44.000000 7.000000 127644.240000

max 92.000000 10.000000 250898.090000

In [154]: bankchurn.iloc[100:150,4:8] # FIRST ROW INDEX

# SECOND COLUMN INDEX # SUBSETTING DATA FRAME

Out[154]:
Geography Gender Age Tenure

100 France Female 40 6

101 France Female 44 6

102 France Male 31 9

103 Spain Male 36 7

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 6 of 19
Day 1 16/12/19, 17:40

104 Spain Female 65 1

105 Spain Female 46 4

106 Germany Male 32 1

107 Germany Female 36 2

108 Spain Male 33 5

109 Germany Male 35 9

110 Germany Male 30 3

111 Germany Male 39 7

112 France Male 42 2

113 Spain Male 36 9

114 Germany Male 28 9

115 Germany Female 30 9

116 Germany Female 37 6

117 France Female 41 1

118 Germany Female 31 8

119 Germany Male 34 4

120 France Male 34 8

121 Spain Male 39 6

122 France Female 39 6

123 Germany Female 48 10

124 France Female 28 3

125 France Male 42 9

126 France Female 52 1

127 Germany Male 56 0

128 France Female 41 7

129 France Male 24 9

130 France Female 34 3

131 Germany Female 33 9

132 France Male 38 9

133 France Male 25 1

134 Germany Male 39 7

135 Germany Female 50 5

136 Germany Female 34 5

137 France Male 40 2

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 7 of 19
Day 1 16/12/19, 17:40

138 Spain Female 48 2

139 Spain Female 35 1

140 Germany Male 44 10

141 France Male 34 7

142 France Female 43 5

143 Spain Male 52 2

144 France Female 31 5

145 Spain Female 21 5

146 Spain Female 29 8

147 France Male 37 5

148 France Male 44 9

149 France Male 32 0

In [155]: # INDIVIDUAL STATISTICAL FUNCTIONS - NUMPY

print(np.min(bankchurn.Balance))
print(np.max(bankchurn.Balance))
print(np.mean(bankchurn.Balance))
print(np.median(bankchurn.Balance))
print(np.var(bankchurn.Balance))
print(np.std(bankchurn.Balance))
print(np.quantile(bankchurn.Balance,0.25))
print(np.quantile(bankchurn.Balance,0.50))
print(np.quantile(bankchurn.Balance,0.75))
print(np.quantile(bankchurn.Balance,0.80)) # 80%
print(bankchurn.Balance.skew()) # Pandas
print(bankchurn.Balance.kurt()) # Pandas

0.0
250898.09
76485.88928799961
97198.54000000001
3893046832.3731775
62394.285254125454
0.0
97198.54000000001
127644.24
133710.358
-0.14110871094154384
-1.489411767941925

In [156]: print(np.quantile(bankchurn.Balance,0.50))# Q2
print(np.percentile(bankchurn.Balance,50))# Q2

97198.54000000001
97198.54000000001

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 8 of 19
Day 1 16/12/19, 17:40

In [157]: bankchurn.dtypes

Out[157]: RowNumber int64

CustomerId int64
Surname object
CreditScore int64
Geography object
Gender object
Age int64
Tenure int64
Balance float64
NumOfProducts int64
HasCrCard int64
IsActiveMember int64
EstimatedSalary float64
Exited int64
dtype: object

In [158]: # NON NUMERIC DATA - object

# SOME NUMERIC COLUMNS MAY BE CATEGORICAL IN
# NATURE - 0 & 1 (No- 0 & Yes- 1)

In [159]: pd.value_counts(bankchurn.HasCrCard)

Out[159]: 1 7055
0 2945
Name: HasCrCard, dtype: int64

In [160]: pd.value_counts(bankchurn.Exited)
# 1 - Customer Exited 0 - Customer Not Exited

Out[160]: 0 7963
1 2037
Name: Exited, dtype: int64

In [161]: pd.value_counts(bankchurn.IsActiveMember)

Out[161]: 1 5151
0 4849
Name: IsActiveMember, dtype: int64

In [162]: bankchurn.columns

Out[162]: Index(['RowNumber', 'CustomerId', 'Surname', 'CreditScore', 'Geogr

aphy',
'Gender', 'Age', 'Tenure', 'Balance', 'NumOfProducts', 'Has
CrCard',
'IsActiveMember', 'EstimatedSalary', 'Exited'],
dtype='object')

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 9 of 19
Day 1 16/12/19, 17:40

In [163]: categorycols=['HasCrCard','IsActiveMember',
'Exited','Geography','Gender']

In [164]: for col in categorycols:

freqcounts=pd.value_counts(bankchurn[col])
print(freqcounts)

1 7055
0 2945
Name: HasCrCard, dtype: int64
1 5151
0 4849
Name: IsActiveMember, dtype: int64
0 7963
1 2037
Name: Exited, dtype: int64
France 5014
Germany 2509
Spain 2477
Name: Geography, dtype: int64
Male 5457
Female 4543
Name: Gender, dtype: int64

In [165]: # DATA VIZUALIZATION

pd.value_counts(bankchurn.Gender).plot(
kind='pie')

Out[165]: <matplotlib.axes._subplots.AxesSubplot at 0x1a2adf6610>

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 10 of 19
Day 1 16/12/19, 17:40

In [166]: # ADVANCE PLOTS - HISTOGRAM, BOXPLOT & DENSITY

# CURVE - UNIVARIATE PLOTS
# 3 PLOTS IDENTIFY SKEWNESS, KURTOSIS, OUTLIERS,
# MISSING VALUES, HOW CLOSE TO NORMAL DISTRIBUTION
# HISTOGRAM - FREQUENCY DISTRIBUTION TABLE
# BOXPLOT - QUARTILES ARE USED
# DENSITY CURVE - Z SCORE OR STANDARD SCORE

# BOXPLOT OUTLIERS DETECTTION - IQR=Q3-Q1

# Q1-1.5*IQR
# Q3+1.5*IQR

In [167]: bankchurn.Balance.plot(kind='hist',color="red")

Out[167]: <matplotlib.axes._subplots.AxesSubplot at 0x1a2d146110>

In [168]: bankchurn.EstimatedSalary.plot(kind='box',vert=False)

Out[168]: <matplotlib.axes._subplots.AxesSubplot at 0x1a2ecde950>

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 11 of 19
Day 1 16/12/19, 17:40

In [169]: bankchurn.Balance.plot(kind='density')

Out[169]: <matplotlib.axes._subplots.AxesSubplot at 0x1a3028f4d0>

In [170]: # BIVARIATE STATISTICS

# COVARIANCE & CORRELATION
bankchurn[['Balance','EstimatedSalary']].cov()
# Covariance

Out[170]:
Balance EstimatedSalary

Balance 3.893436e+09 4.592389e+07

EstimatedSalary 4.592389e+07 3.307457e+09

In [171]: bankchurn[['Balance','EstimatedSalary']].corr()
# Correlation

Out[171]:
Balance EstimatedSalary

Balance 1.000000 0.012797

EstimatedSalary 0.012797 1.000000

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 12 of 19
Day 1 16/12/19, 17:40

In [172]: bankchurn.plot(kind='scatter',x='Balance',
y='EstimatedSalary')

Out[172]: <matplotlib.axes._subplots.AxesSubplot at 0x1a2d9c6990>

In [173]: # GROUPING FUNCTION - "groupby" - Slicing Function

# LEFTSIDE GROUPBY MUST BE NUMERICAL
# RIGHTSIDE GROUPBY MUST BE CATEGORICAL
# SPECIFY A STATISTICAL FUNCTION MUST
# 2 TYPES CUSTOMER - EXITED & NOT EXITED
# IS THE AVERAGE BALANCE MAINTAINED BY THESE 2 TYPE
# OF CUSTOMERS EQUAL/SAME.

print(pd.value_counts(bankchurn.Exited))
bankchurn.Balance.groupby(bankchurn.Exited).mean()

0 7963
1 2037
Name: Exited, dtype: int64

Out[173]: Exited
0 72745.296779
1 91108.539337
Name: Balance, dtype: float64

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 13 of 19
Day 1 16/12/19, 17:40

In [174]: pd.crosstab(bankchurn.Exited,bankchurn.Gender,
normalize='index')
# CROSS TABULATION _ FREQUENCY of 2 Categorical
# Variables

Out[174]:
Gender Female Male

Exited

0 0.427477 0.572523

1 0.559156 0.440844

In [175]: bankchurn.Balance.groupby([bankchurn.Exited,
bankchurn.Gender,
bankchurn.Geography]).mean()

Out[175]: Exited Gender

Geography
0 Female
France 58424.310061
Germany 118828.514362
Spain 56594.820688
Male France 61774.503758
Germany 119896.529105
Spain 61871.626285
1 Female France 67755.162630
Germany 119673.872321
Spain 71997.673680
Male France 75710.827800
Germany 121202.242432
Spain 73167.867802
Name: Balance, dtype: float64

In [176]: bankchurn.Balance.groupby(
bankchurn.Exited).mean()

Out[176]: Exited
0 72745.296779
1 91108.539337
Name: Balance, dtype: float64

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 14 of 19
Day 1 16/12/19, 17:40

In [177]: # HYPOTHESIS TESTING - COMPARISION OF MEANS

# COMPARING 2 GROUPS - 2 SAMPLE INDEPENDENT TTEST
# COMPARING MORE THAN 2 GROUPS - ANOVA SINGLE
# FACTOR
# ASSUMPTIONS:
# MEANS OF GROUPS MUST BE DIFFERENT
# NUMERICAL VARIABLE MUST BE CONTINOUS, CLOSER
# TO NORMAL DISTRIBUTION, NO OUTLIERS & MISSING
# VALUES
# OTHER VARIABLE MUST BE CATEGORICAL WITH
# EXACTLY 2 GROUPS OR MORE THAN 2 GROUPS

#NULL - THERE IS NO SIGNIFICANT DIFFERENCE IN

# AVERAGE OF GROUP 1 & GROUP 2
#ALT -THERE IS SIGNIFICANT DIFFERENCE IN
# AVERAGE OF GROUP 1 & GROUP 2

# P-VALUE LESS THAN 0.05, REJECT NULL & ACCEPT

# ALTERNATE
# P-VALUE GREATER THAN 0.05, FAIL TO REJECT NULL
# (ACCEPT NULL) & REJECT ALTERNATE

# PROCESS IN PYTHON
# GROUPBY - TO IDENTIFY NUM OF GROUPS & CONFIRM
# MEAN DIFFERENCE
# SPLIT THE DATAFRAME INTO SUBSETS BASED ON
# GROUPS
# CONDUCT THE RELEVANT TEST & INFER BASED ON
# P-VALUE

In [178]: # SPLIT DATAFRAME INTO EXITED AND NONEXITED

exited=bankchurn[bankchurn.Exited==1]
nonexited=bankchurn[bankchurn.Exited==0]
print(exited.shape)
print(nonexited.shape)

(2037, 14)
(7963, 14)

In [179]: from scipy.stats import ttest_ind

# 2 Sample Independent T test
ttest_ind(exited.Balance,nonexited.Balance,
equal_var=False)
# SINCE p-value less tha 0.05, REJECT NULL

Out[179]: Ttest_indResult(statistic=12.47128032005069, pvalue=6.318663518527

793e-35)

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 15 of 19
Day 1 16/12/19, 17:40

In [180]: # IS THE AVERAGE BALANCE MAINTAINED BY MALE &

# FEMALE CUSTOMERS SAME or EQUAL?
# groupby
# split data into male & female
# Frame NULL & ALT
# Conduct Test & Infer

In [181]: bankchurn.Balance.groupby(
bankchurn.Gender).mean()

Out[181]: Gender
Female 75659.369139
Male 77173.974506
Name: Balance, dtype: float64

In [182]: # NULL - THERE IS NO SIGNIFICANT DIFFERENCE IN

# AVERAGE BALANCE OF MALE & FEMALE CUSTOMERS
# ALT - THERE IS SIGNIFICANT DIFFERENCE IN
# AVERAGE BALANCE OF MALE & FEMALE CUSTOMERS
male=bankchurn[bankchurn.Gender=='Male']
female=bankchurn[bankchurn.Gender=='Female']

In [183]: ttest_ind(male.Balance,female.Balance,
equal_var=False)
# SINCE p-value greater than 0.05, Fail to Reject
# Null

Out[183]: Ttest_indResult(statistic=1.2095754533033731, pvalue=0.22647131768

566747)

In [185]: # IS THE AVERAGE BALANCE MAINTAINED BY DIFFERENT

# GEOGRAPHY EQUAL OR SAME?
bankchurn.Balance.groupby(
bankchurn.Geography).mean()

Out[185]: Geography
France 62092.636516
Germany 119730.116134
Spain 61818.147763
Name: Balance, dtype: float64

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 16 of 19
Day 1 16/12/19, 17:40

In [188]: # SINCE MORE THAN 2 LEVELs - ANOVA ONE WAY or

# ANOVA SINGLE FACTOR
france=bankchurn[bankchurn.Geography=='France']
germany=bankchurn[bankchurn.Geography=='Germany']
spain=bankchurn[bankchurn.Geography=='Spain']

# NULL - THERE IS NO SIGNIFICANT DIFFERENCE IN

# AVERAGE BALANCE OF FRANCE, GERMANY & SPAIN
# alt - THERE IS SIGNIFICANT DIFFERENCE IN
# AVERAGE BALANCE OF FRANCE, GERMANY & SPAIN

In [189]: from scipy.stats import f_oneway

f_oneway(france.Balance,germany.Balance,
spain.Balance)
# SINCE p-value less than 0.05, REJECT NULL

Out[189]: F_onewayResult(statistic=958.4254463368385, pvalue=0.0)

In [191]: # DISTRIBUTIONS
# T TEST - STUDENT T DISTRIBUTION
# ANOVA SINGLE FACTOR - F DISTRIBUTION
# CHI SQUARE TEST - CHI DISTRIBUTION

# BERNOULLI BINOMIAL DISTRIBUTION - BINARY

# YES/NO or TRUE/FALSE - LOGISTIC REGRESSION
# POISSON DISTRIBUTION - INSURANCE DATA
# EXPONENTIAL DISTRIBUTION - TRAFFIC DATA
# NORMAL DISTRIBUTION or GAUSSIAN DISTRIBUTION

In [193]: # ONE NUMERIC VARIABLE OTHER CATEGORICAL VARIABLE

# WITH EXCATLY 2 LEVELS - T TEST
# ONE NUMERIC VARIABLE OTHER CATEGORICAL VARIABLE
# WITH morethan 2 LEVELS - ANOVA SINGLE FACTOR

# BOTH THE VARIABLES NON NUMERICAL or Categorical

# CHI SQUARE TEST OF INDEPENDENCE
# INPUT for CHI SQUARE TEST IS CROSSTABULATION
# NULL - THERE IS NO ASSOCIATION/RELATIONSHIP
# between BOTH VARIABLES
# ALT - THERE IS ASSOCIATION/RELATIONSHIP
# between BOTH VARIABLES

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 17 of 19
Day 1 16/12/19, 17:40

In [194]: # IS THERE ASSOCIATION BETWEEN GENDER & EXITED?

pd.crosstab(bankchurn.Gender,bankchurn.Exited)

Out[194]:
Exited 0 1

Gender

Female 3404 1139

Male 4559 898

In [197]: from scipy.stats import chi2_contingency

chi2_contingency(pd.crosstab(bankchurn.Gender,
bankchurn.Exited))
# Since 2.2482100097131755e-26 less than 0.05, REJECT
# NULL

Out[197]: (112.91857062096116, 2.2482100097131755e-26, 1, array([[3617.5909,

925.4091],
[4345.4091, 1111.5909]]))

In [198]: # IS THERE ASSOCIATION BETWEEN GENDER & HASCRCARD

chi2_contingency(pd.crosstab(bankchurn.Gender,
bankchurn.HasCrCard))
# SINCE p-value greater than 0.05, Fail to REJECT NULL

Out[198]: (0.30756077917984026, 0.5791808600631774, 1, array([[1337.9135, 32

05.0865],
[1607.0865, 3849.9135]]))

In [200]: # PAIRED SAMPLE T TEST - BEFORE & AFTER

# INDIA's INFLATION BEFORE & AFTER DEMONITIZATION
befordemon=[6.59,6.13,6.46,5.30,4.14,3.35]
afterdemon=[2.23,1.86,2.62,2.61,2.21,1.09]
print(np.mean(befordemon))
print(np.mean(afterdemon))

# NULL - THERE IS NO SIGNIFICANT DIFFERENCE IN AVERAGE

# INFLATION BEFORE AND AFTER DEMONITIZATION
# ALT - THERE IS SIGNIFICANT DIFFERENCE IN AVERAGE
# INFLATION BEFORE AND AFTER DEMONITIZATION

5.328333333333334
2.1033333333333335

In [202]: from scipy.stats import ttest_rel

ttest_rel(befordemon,afterdemon)
# SINCE p-value less than 0.05, REJECT NULL

Out[202]: Ttest_relResult(statistic=7.429383454110612, pvalue=0.000696190145

3565623)

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 18 of 19
Day 1 16/12/19, 17:40

In [203]: # BEFORE & AFTER GST MONTHLY INFLATION

beforegst=[1.86,2.62,2.61,2.21,1.09,1.08]
aftergst=[2.52,2.89,3.24,3.97,4.00,5.11]
print(np.mean(beforegst))
print(np.mean(aftergst))

1.9116666666666668
3.6216666666666666

In [204]: ttest_rel(beforegst,aftergst)

Out[204]: Ttest_relResult(statistic=-2.8027263826421933, pvalue=0.0378716146

2210822)

In [ ]:

http://localhost:8889/nbconvert/html/SBIIT%20B2/Day%201%20.ipynb?download=false Page 19 of 19