Correlation Analysis
Correlation analysis is an analysis of the relationship of two or
more variables.
Types of correlation:
1. Positive or Negative Correlation
2. Simple, partial and multiple correlation
3. Linear and non linear correlation
Methods of correlation Analysis :
1. Graphic method
a. Scatter Diagram method
b. Graphic plot
2. Statistical method
a. Karl Pearsons Coefficient of Correlation
b. Rank Method
c. Concurrent Deviation Method
d. Method of Least Squares
Problem 1:
Draw a correlation graph
Months Jan
Feb
Income 1000
1200
Expens 900
1200
es
from the
March
1400
1300
following data:
April
May
1800
1900
1600
1700
2500
2000
1500
1000
500
0
Income
Expenses
June
2000
1900
July
2200
2000
Inference : There exists a very close positive correlation between
income and expenses
Statistical Method
1. Karl Pearsons coefficient of Correlation:
a. Arithmetic Mean Method
r = xy/ x 2 . y 2
where x=X -
and y=Y- Y
Problem2 :
Compute coefficient of correlation for the following data through
Karl Pearsons coefficient method
X
Y
25
20
Find mean X
35
15
45
10
52
14
= X / N
20
23
33
18
40
22
30
30
580
420
600
550
= Y /N
x=X - X
y=Y- Y
Use formula above to calculate
Problem 3 : Assumed Mean method
Calculate Karl Pearsons coefficient of correlation
Income 230
560
490
360
270
480
Expens 200
440
350
250
240
300
es
d=x-A
r=
dxdy( dx . dy )/N
dx 2 ( dx ) 2/ N . dy 2( dy ) 2 /N
Karl Pearsons coefficient for Grouped data
r=
f dxdy(fdx . fdy)/ N
f dx 2( fdx ) 2/ N . f dy 2( fdy ) 2 /N
Problem 4:
Calculate coefficient of correlation for the following data
Weekly
Weekly Expenses
120140160180200Income 100-120
125150
150175
175200
200225
225250
140
160
180
200
220
2
1
1
5
Total
12
9
12
10
11
8
10
50
Rank Correlation Co-efficient:
When ranks are given
1. Find out the difference of the two ranks( i.e., D) for the two
variables
2. Take the squares of these differences ( i.e, D 2) and find D2
3. Substitute the values in the formula
rR = 1 - 6D2/ N3 N
Problem : 5
Calculate rank co-efficient of correlation for 12 students in 2
different subjects
Studen No.
ts
Subjec I
10
10 11 1
2
11 12 2
t
Subjec
t
II
12
11
Regression Analysis:
Regression is a statistical technique, through which estimation of
unknown variable from the known can be done.