Você está na página 1de 10

APPLIED BIOSTATISTICS FOR BIOTECHNOLOGISTS ASSIGNMENT

TOPIC :- REGRESSION

NAME:- TEOTIA NIDHI


CLASS:- M.Sc BIOTECHNOLOGY
CONTENTS

INTRODUCTION
• REGRESSION
• REGRESSION LINES
• REGRESSION COEFFICIENT

TYPES OF REGRESSION
METHODS OF STUDYING
REGRESSION
PROPERTIES OF REGRESSION
SOLVED EXAMPLES
INTRODUCTION

REGRESSION is used to denote estimation or prediction of the average value


of one variable for a specified value of the other variables. One of the variable is
called independent or the explained variable or predictors and the other is
called dependent or the explaining variable.
“Regression is the measure of the average relationship between two or more
variables in terms of the original unit of data” M.M Blair
The estimation and prediction is done by means of suitable equation derived on
the basis of available bivariate data. Such an equation is known as regression
equation and its geometrical representation is called regression curve.
1) Regression Equation of X on Y :-
X - X́ = b xy (Y- Ý )
σx
= r σy ( Y- Ý ) [ It estimates X for given value of Y ]

2) Regression Equation Y on X :-
Y- Ý = b yx (X- X́ )
σy
=r σx (X- X́ ) [It estimate Y for a given value of X]

Where X=Value of x
X́ =Mean of x
σ x = Standard deviation of x series
r = Correlation coefficient
Y= Value of Y
Ý = Mean of Y
σy = Standard deviation of y series
b = Slope or coefficient of regression

Regression Lines :-
If a bivariate data are plotted as points on graph paper , it will be found
that the concentration point follows a certain patterns showing the
relationship between the variables. When the trend points are found to be
linear, we determine the best fitting straight line by Least Square Method.
Such straight lines which are used to obtain best estimates of one
variables for given value of the other ,are called regression lines.
If two variables are linearly related ,then relation can be expressed as
Y=bx = a. where b =slope of the line relating Y to X and ‘a’ is the ‘Y’
intercept of the line.
A line regression is the straight line which gives the best fit in the least
square sense to given sets of data.

Regression coefficient:-
1) The regression coefficient (b) is an expression of how much ( on the
average) one dependent variable (Y) may be expected to change per
unit change in some other independent variable (X).
2) It is denoted by letter ‘b’.
3) The regression coefficient of Y on X =
σy S . D .of Yseries
= byx = r σx S . D . of X series
.
The regression coefficient of X on Y =
σx S . D . of X series
= bxy = r σy S . D . of Y series
.

Types of regression:-
a) SIMPLE REGRESSION :-
I. Here the dependent variable (criterion) is a function of single
independent variable (predictor) .
II. The score of the dependent variable is predicted from the given
scores of single predictor.
EXAMPLE : Height of person on his weight
Simple liner regression model-
The regression model describes the mean of that normally distributed dependent
variable Y as a function of the predictor or independent variable X:
Yi = β o + β 1 X i + εi
It is simple because it contains only 1 independent variables. It is linear because
the independent variables appears only in the first power, if we graph the mean
1.
of Y vs X , the graph is a straight line with intercept β 0 and slope ¿
β¿

The scatter diagram is a useful diagnostic tool for checking out the validity of
features of the simple linear regression model.
Testing for independence –
In addition to being able to predict the mean at various levels of the independent
variables , regression data can also used to test for the independence between
the two variables under investigation. Such a statistical test can be viewed in
two ways: through the coefficient of correlation or through the slope.
1) The correlation coefficient r measures the strength of the relation between
two variables it is an estimate of an unknown population correlation
coefficient ρ (rho),the same way the sample mean x́ is used as an
estimate of an unknown population mean µ.
t= r √ (n−2)/(1−r 2)
the procedure is often performed as two sided ,i.e
H A : ρ≠0
And it is t test with n-2 degrees of freedom
2) The role of the slope β 1 ,since the regression model describes the mean
of the dependent variable Y as a function of the predictor or the
1
independent variable X, µ y = β0 = ¿ x
β¿
If β 1 =0 ,Y and X would be independent. The test for H 0 : β1 = 0.

b) MULTIPLE REGRESSION:-
i. Here the dependent variable is a function of two or more predictors
ii. The scores are predicted from the scores of more than one predictors
iii. It may be linear or non linear.
EXAMPLE : Thyroid calcitonin on combination of thyroxine secretion
and serum calcium.
Regression model with several independent variables-
Suppose that we want to consider k independent variables
simultaneously-
k
µ i=β 0=∑ β j x ji
j=1

The model above is referred to as multiple linear regression model.


Effective modification -consider a multiple regression model involving
two independent variables :

Polynomial regression:-
Consider the multiple regression model involving one independent
variable:-

Testing hypothesis in multiple regression:-


Test for a single variable- let us assume that we now wish to test whether the
addition of one particular independent variable of interest adds significantly

C) LINEAR REGRESSION-
i. here the dependent variable is linearly correlated with the predictor (independent
variable)
ii. the scores of dependent variables are predicted by working out an equation for a
straight line, depending on the linear association between the two.
The statistical analysis is to find out the exact position of the straight line is known as
linear regression analysis.
ii. Its equation is y = a = bx
iii. The slope of the line b in the equation is known as the regression coefficient
it shows that y changes b times as fast as x.
iv. Symbolically the regression coefficient of y on x is b yx .
(B) i. If the line of regression is so chosen that the sum of square of deviation
parallel to the axis of x is minimized fig 14.2 (b), it is called the line of
regression of X on Y and it gives the best estimate of x for any value of y.
ii. The regression equation in this case is x = a = by.
Computation of linear equation

(d) NON LINEAR REGRESSION


If the criterion has a non linear correlation with the predictor, the score of the criterion have
to be predicted in terms of a curved line like a sigmoid or hyperbolic or exponential curve,
according to their form of association.
Method of studying regression-
1)GRAPHIC METHOD-
i. the points are platted on a graph paper representing pairs of value of concerned variables
ii. independent variables are taken on the horizontal axis & dependent on vertical axis.
iii. these points give a picture of a scatter diagram
2)ALGRBEAIC METHOD-
i. a regression line is a straight line fitted to the data by the method of least squares
ii. it indicates the best possible mean value of one variable corresponding to the mean
value of the other.
iii. there are always two regression lines constructed for the relationship between the two
variables viz .X &Y.

Properties of regression –
1) It is an expression of the dependent variable as a function of the independent
variable.
2) Regression predicts only a probable score of the criterion on a given score of the
predicator.
3) It is worked out using a statistic called the regression coefficient.

Solved examples

Example 2 The correlation coefficient between X &Y is 0.60. if the variance of x


=225,the variance of Y= 400,means of X =10 & means of Y=20,find the equation
of the regression lines of (i) Y on X and (ii) X on Y.
SOLUTION :
2
Variance of X i.e., σ x = 225 σ x = √ 225 = 15
Variance of Y i.e., σ 2x =400 σ y =√ 400=20

r= 0.6 X́ =¿ 10 Ý =¿ 20
σy
byx= r σx = 0.6 × 20/15 = 0.8

15
bxy=o.6 × 20 =0.45
(i) Regression equation Y on X
Y - Ý =byx ( X − X́ )
Y-20=0.8 (X-10)
Y= 0.8X-8+20
Y = 0.8 X + 12

(ii) Regression equation of X on Y


X - X́ =bxy (Y −Ý )
X-10 = 0.45 (Y-20)
X= 0.45Y-9
X= 0.454-9=10=0.45Y+1

Reference

Introduction to Biostatistics: “A textbook of biometry” by Pranab Kumar


Banerjee
Introductory Biostatistics by John Wiley & Sons.

Você também pode gostar