Click to learn more about author Kartik Patel.

What Is Naïve Bayes Classification?

Naive Bayes is a classification algorithm that is suitable for binary and multiclass classification. It is a supervised classification technique used to classify future objects by assigning class labels to instances/records using conditional probability. In supervised classification, training data is already labeled with a class. For example, if fraudulent transactions are already flagged in transactional data and if we want to classify future transactions into fraudulent/non-fraudulent, then that type of classification would be called supervised.

Let’s say we want to classify fruit. Fruit may be considered to be an apple if it is red, round, and about 3″ in diameter. If we have data on 1000 pieces of fruit, including features or characteristics of each fruit, we can classify the 1000 pieces of fruit characteristics such as shape, length, color, sweetness, sourness, etc.

When we look at the table above, we see that 50 percent of the fruit is bananas, 30 percent is oranges, and 20 percent is other types of fruit.

The Naive Bayes classifier assumes that every feature/predictor is independent, which is not always the case, so it is important to understand the type of data you are analyzing before choosing this or any other analytical technique.

In order to make the best use of the Naïve Bayes method, the training dataset should be adequate enough to represent the entire population — containing every combination of class label and attributes. Naïve Bayes performs well in cases of categorical input variables compared to numerical variables. For numerical variables, normal distribution is assumed, which is a strong assumption.

How Can Naïve Bayes Be Used for Enterprise Analysis?

This technique can be useful in evaluating many applications.

• Weather Forecasting: Based on temperature, humidity, pressure, etc., an organization can predict if it will be rainy/sunny/windy tomorrow.
• Fraud Analysis: Based on various bills submitted by an employee for reimbursement for expenditures on food, travel, etc., a business can predict the likelihood of fraud.

Use Case 1

Business Problem: A bank loan officer wants to predict if a loan applicant will be a bank defaulter or non-defaulter based on attributes such as loan amount, monthly installment, employment tenure, the number of times delinquent, annual income, debt to income ratio, etc. Here the target variable would be “past default status,” and the predicted class would contain the values “yes or no,” representing the “likely to default/unlikely to default” class, respectively.

Business Benefit: Once classes are assigned, the bank will have a loan applicant dataset with each applicant labeled as “likely/unlikely to default.” Based on these labels, the bank can easily make a decision on whether to give a loan to an applicant and how much credit and interest rate each applicant is eligible to receive.

Use Case 2

Business Problem: A doctor wants to predict the likelihood of successful treatment of a patient disease or condition based on various attributes of a patient such as blood pressure, hemoglobin level, blood sugar level, the name of a drug given to the patient, the type of treatment given to the patient, etc. Here the target variable would be “past cure status,” and the predicted class would contain values “yes or no,” meaning “prone to cure/not prone to cure,” respectively.

Business Benefit: Given the health and body profile of a patient and recent treatments and drugs administered, the probability of a cure can be predicted, and changes in treatment and drug recommendations can be suggested if required.

The Naive Bayes is a classification algorithm that is suitable for binary and multiclass classification. Naïve Bayes performs well in cases of categorical input variables compared to numerical variables. It is useful for making predictions and forecasting data based on historical results.