Write a formulation for Exact Bayesian classifier for classifying a p-dimensional record (x(i)) into a class ( k c ) giv
Posted: Mon May 09, 2022 12:38 pm
Write a formulation for Exact Bayesian classifier for
classifying a p-dimensional record (x(i)) into a class ( k c )
given the training data set { } 1 ( ), ( ) n i D i yi = = x and yi
c c c ( ) , ,..., ∈{ 1 2 m} .
a. Modifying the Exact Bayesian classifier assuming that
predictors are independent to arrive at Naïve Bayesian classifier.
Explain why these assumptions may not adversely affect the
classification performance of the Naïve Bayesian classifier is
relative to the classification performance of the Exact Bayesian
classifier.
b. If the dimensionality of x is p = 10, and each predictor X j
(j=1, 2, …, p) has three categories, estimate the minimum number of
records (nmin_ebc) required to cover all possible unique
combinations of predictor values for an Exact Bayesian classifier.
In this case, what would be the minimum number of records
(nmin_nbc) required for a Naïve Bayes classifier?
classifying a p-dimensional record (x(i)) into a class ( k c )
given the training data set { } 1 ( ), ( ) n i D i yi = = x and yi
c c c ( ) , ,..., ∈{ 1 2 m} .
a. Modifying the Exact Bayesian classifier assuming that
predictors are independent to arrive at Naïve Bayesian classifier.
Explain why these assumptions may not adversely affect the
classification performance of the Naïve Bayesian classifier is
relative to the classification performance of the Exact Bayesian
classifier.
b. If the dimensionality of x is p = 10, and each predictor X j
(j=1, 2, …, p) has three categories, estimate the minimum number of
records (nmin_ebc) required to cover all possible unique
combinations of predictor values for an Exact Bayesian classifier.
In this case, what would be the minimum number of records
(nmin_nbc) required for a Naïve Bayes classifier?