Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

Naive Bayes Example

created by André Rodrigues
Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

Naı̈ve Bayes Classifier

Simplest probabilistic model for classification:

d
Y
P (X | Y = y) = P (Xi | Y = y)
i=1

where X is a set of attributes and Y is the class.

Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

Naı̈ve Bayes Classifier

Naive Bayes assumes that attributes X are independent given

the class variable Y.
Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

Naı̈ve Bayes Classifier

To classify a test example, the Naive Bayes classifier computes

the posterior probability for each class Y:
d
Y
P (Y | X) = P (Y ) P (Xi | Y = y)
i=1
Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

Given a set of observations...

Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

...build the Naı̈ve Bayes Classifier and...

Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

...Conditional Probability Tables (CPTs or

network parameters)
Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

...Conditional Probability Tables (CPTs or

network parameters)
Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning Probabilistic Reasoning

Now, use the model to predict classes

Given a new patient with symptoms: Runny nose=no,

Headache=no, Chills=true, Fever=true
Calculate:

P Y es = P (F lu = Y es | RN = no, H = no, Chills = true, F ever = tru

P N o = P (F lu = N o | RN = no, H = no, Chills = true, F ever = true)

If P Y es > P N o, patient has Flu, otherwise, patient has no Flu.