Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Machine Learning Algorithms and its application in real life

A machine learning algorithm is a piece of software (math or program logic) that allows experts to
examine, analyze, interpret, and explore enormous, complicated datasets. Each algorithm follows a set of
instructions to achieve the goal of producing predictions or classifying information by learning,
establishing, and identifying patterns in the data. Machine learning algorithms define the rules and
methods that a system should follow while tackling a certain task.
Algorithms evaluate and simulate data in order to anticipate the outcome within a given range.
Furthermore, when fresh data is introduced into these algorithms, they learn, optimize, and improve
depending on feedback from past predictions. Simply put, machine learning algorithms get ‘smarter' with
each iteration.
Based on the learning approaches, machine learning algorithms are grouped into four types: supervised,
semi-supervised, unsupervised, and reinforcement learning. For forecasting values, recognizing
similarities, and uncovering odd data patterns, regression and classification algorithms are the most
common solutions.
Decision trees
A decision tree allows you to view the map of potential outcomes for a sequence of decisions. It helps
businesses to analyze potential outcomes and then make an informed decision based on characteristics
such as benefits and probabilities that are advantageous to them.
Decision tree algorithms have the capacity to predict the optimal option based on a mathematical
architecture and may also be useful when brainstorming about a specific decision. Each conclusion can
generate child nodes, which can lead to new possibilities. The technique creates a tree-like structure that
may be utilized to solve classification challenges. Consider the decision tree below, which assists in
finalizing a weekend plan based on the weather prediction.
Naive Bayes algorithm

Naive Bayes is a probabilistic machine learning technique that is used to solve classification issues and is
based on the Bayesian probability model. The algorithm's core premise is that the characteristics under
consideration are independent of one another, and that a change in the value of one does not affect the
value of the other. Consider a ball, such as a cricket ball, if it is red, spherical, has a diameter of 7.1-7.26
cm, and weighs 156-163 g. Although all of these characteristics may be interconnected, each one
contributes to the likelihood that it is a cricket ball. Therefore the algorithm is referred described as
"naive."

KNN classification algorithm

It keeps all of the existing use cases and classifies new use cases (or data points) by categorizing them.
This categorization is based on the similarity score of the most recent use cases to the ones that are
currently accessible is a supervised machine learning technique, where 'K' refers to the number of nearby
points taken into account for classifying and separating the known n groups. The algorithm learns at each
step and iteration; therefore no particular learning phase is required. The categorization is determined by
the majority vote of the neighbor.

Random forest algorithm

To solve classification and regression issues, random forest techniques employ numerous decision trees.
It is a supervised machine learning approach in which distinct decision trees are created during training
on different data. These methods aid in the estimation of missing data and tend to maintain accuracy
when a big chunk of data is missing from a dataset.
Recurrent neural networks (RNNs)

Recurrent neural networks are an ANN type that analyzes sequential data. In this case, the previous
step's output serves as the input to the next step. This is made possible by the hidden state, which
remembers information about a sequence. It functions as a memory, retaining information about past
calculations. RNN memory minimizes the neural network's total complexity.

Data mining using decision tree algorithm

One of the many data mining approaches available is decision trees. Decision trees, which may be used
for categorization and prediction, are a technique to help in decision making (Lee et al., 2007). A decision
tree has previously been used for data analysis in various application sectors due to its ability to
effectively classify data and generate successful predictions. We aim to present an overview of the
applications that decision trees can enable in this paper. We specialize on corporate management,
engineering, and health-care administration.

Random forest algorithm on Credit Card Fraud Detection

Credit card firms should detect fraudulent credit card purchases so that customers are not charged for
items they did not buy. However, it is a challenging process since there may be only 1000 occurrences of
fraud in over a million transactions, representing just 0.1 percent of the dataset, resulting in extremely
skewed datasets. The program could predict almost all situations belonging to the majority class, yielding
a high accuracy score. For this, we may still utilize the Random Forest Classifier.
Naive Bayes algorithm in Gmail predicting Spam mails

The Nave Bayes Classifier, which is based on the Bayesian theorem, is a basic probabilistic classifier with
strong independent assumptions. It assumes that the presence or absence of one characteristic has no
bearing on the presence or absence of another. An apple is a fruit that is red in color, round, and roughly 4
inches in diameter. Even though the existence of one of these qualities is dependent on the existence of
another, the classifier considers all of these properties to contribute to the fruit being an apple. Nave Bayes
is a machine learning model that is used for big amounts of data; even if you are dealing with data that
contains millions of records, Nave Bayes is the preferred strategy.

You might also like