Discriminant Analysis

Discriminant Analysis
MBA-BM (2022-24, Sec-A&B)
Prof. Nirali Shah

E-mail: [email protected]
Discriminant analysis
• Discriminant analysis is a technique for analyzing data when

the criterion or dependent variable is categorical in nature the
predictor or independent variables are interval in nature
• Examples of categorical scaled

• Choice of brands (Brand 1, Brand 2, Brand 3)
• Customer satisfaction (satisfied customer, dissatisfied customer)
• Credit worthiness (low risk customer, high risk customer)
• Type of stores (single outlet, part of chain outlets)
• Model
• D = bo + b1X1 + b2X2 + b3X3 + … + bkXk
• Where,
• D = discriminant score
• b’s = discriminant coefficient or weight
• X’s = predictor or independent variable
Discriminant Analysis
Number of dependent variables One
Number of independent variables Multiple
Nature of the dependent variable Categorical
Nature of the independent variable Metric

Objectives of discriminant analysis
• Developing a discriminant function or determining a linear

combination of predictor/independent variables (IVs) to separate
groups of dependent variables (DV)
• Examining significant differences exist among the groups in light of IVs
• Determining IVs contribute to most of the intergroup differences
• Classifying cases to one of the groups based on the values of the
predictor variables
• Developing procedures for assigning new entrants whose
characteristics are known, but group identity is not known
• Evaluating accuracy of classification
Types of discriminant analysis
Number of
categories
Technique Derive
possessed by the
criterion variable
Two-group
One discriminant
Two categories discriminant
function
analysis
Multiple
Three or more Multiple discriminant
discriminant
categories analysis
function
Usage of discriminant analysis in marketing
• Demographic characteristics and store loyalty

• Psychographic characteristics and price-sensitive
• Personality and new product choices
• Market segments and media habits
• Psychographic characteristics and adventurous experiences
• Friends and food habit
• Environmental conscious citizens and shopping preferences
Statistics associated with discriminant analysis
• Canonical correlation
• Measures the extent of association between the discriminant
scores and the groups.
• A measure of association between the single discriminant function
and the set of dummy variables that define the group membership
• Centroid
• The mean values for the discriminant scores for a particular group
• As many centroids as there are groups, as one for each group
• Classification matrix
• Also called confusion or prediction matrix
• Contains the number of correctly classified and
misclassified cases
• The sum of the diagonal elements divided by the total
number of cases represents the hit ratio.
• Discriminant function coefficients

• (Unstandardized) are the multipliers of variables, when the
variables are in the original units of measurement
• Discriminant scores
• The unstandardized coefficients are multiplied by the values of
the variables.
• These products are summed and added to the constant term to
discriminant scores.
D = bo + b1X1 + b2X2 + b3X3 + … + bkXk
• Eigenvalue
• The ratio of between-group to within-group sums of
squares for each discriminant function
• Large Eigenvalues imply superior functions
• Group means and group standard deviations

• Computed for each predictor for each group
• Pooled within-group correlation matrix

• Computed by averaging the separate covariance matrices for all
the groups
• Standardized discriminant function coefficients

• Used as the multipliers when the variables have been
standardized to a mean of 0 and a variance of 1
• Structure correlations
• Also referred to as discriminant loadings
• Represent the simple correlations between the predictors and
the discriminant function.
• Total correlation matrix

• If the cases are treated as if they were from a single sample and the
correlations computed, a total correlation matrix is obtained.
• Wilks’ 
• For each predictor
• The ratio of within-group sum of squares to the total sum of squares
• Value varies between 0 and 1
• Large values (near 1) indicate group means do not seem to be different
• Small values (near 0) indicate the group means seem to be different
Conducting discriminant analysis
Formulate the problem
Compute the discriminant function coefficients
Determine the significance of the discriminant function
Interpret the results
Assess the validity of the discriminant analysis

• To identify the objectives, the criterion (dependent) variable,

and the predictor (independent) variables
• The criterion variable
• Consists of two or more mutually exclusive and collectively
exhaustive categories
• The predictor variables
• Selection based on a theoretical model/ previous research/
experience of the researcher
• Sample
• One part: the estimation or analysis sample - used for estimation of
discriminant function
• Other part: the holdout or validation sample- reserved for validating
discriminant function
• Double cross-validation
• Repeatedly conduct the validation of the discriminant function
• Split the sample each time into different analysis and validation part
• Two approaches to estimate the discriminant coefficients

• Direct method
• Simultaneously includes all the predictor variables irrespective
of discriminating power
• Appropriate when based on previous research or a theoretical
model
• Stepwise discriminant analysis
• Sequentially add the predictor variables based on their ability
to discriminate
• Appropriate when the researcher wants to select a subset of
the predictors for inclusion in the discriminant function
• To determine the significance of discriminant function
• Null hypothesis:
• The means of all discriminant functions in all groups are
equal can be statistically tested
• If the rejection of the null hypothesis
• Indicates significant discrimination among groups
• Assessing the accuracy of the model

• Using hit ratio
• Relative importance of the variables
• By examining the absolute magnitude
• Creation of discriminant model
• Using the discriminant weights, or coefficients
• Functions at group centroids
• Testing discriminant score against the group means to decide
• Wilk’s Lambda
• Assessing uniqueness among two groups

Discriminant Analysis

Uploaded by

Copyright:

Available Formats

Discriminant Analysis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Discriminant Analysis

Uploaded by

Copyright:

Available Formats

Discriminant Analysis

MBA-BM (2022-24, Sec-A&B)

Prof. Nirali Shah

• Discriminant analysis is a technique for analyzing data when

• Examples of categorical scaled

Number of dependent variables One

Number of independent variables Multiple

Nature of the dependent variable Categorical

Nature of the independent variable Metric

• Developing a discriminant function or determining a linear

• Demographic characteristics and store loyalty

• Discriminant function coefficients

• Group means and group standard deviations

• Pooled within-group correlation matrix

• Standardized discriminant function coefficients

• Total correlation matrix

Formulate the problem

Compute the discriminant function coefficients

Determine the significance of the discriminant function

Interpret the results

Assess the validity of the discriminant analysis

• To identify the objectives, the criterion (dependent) variable,

• Two approaches to estimate the discriminant coefficients

• To determine the significance of discriminant function

• Assessing the accuracy of the model

You might also like