Classification Algorithm in Machine Learning
In this page we will learn Classification Algorithm in Machine Learning, What is the Classification Algorithm in Machine Learning?, Learners in Classification Problems, Types of ML Classification Algorithms, Evaluating a Classification model, Log Loss or CrossEntropy Loss, Use cases of Classification Algorithms.
As we all know, the Regression and Classification Algorithms are two types of Supervised Machine Learning algorithms. We predicted the output for continuous values using Regression techniques, but we require Classification methods to predict categorical values.
What is the Classification Algorithm in Machine Learning?
The Classification algorithm is a Supervised Learning
technique that uses training data to determine the category of
new observations. Classification is the process of a software
learning from a dataset or observations and then classifying
fresh observations into one of several classes or groupings.
Yes or No, 0 or 1, Spam or Not Spam, cat or dog, and so forth.
Targets/labels or categories are all terms that can be used to
describe classes.
Unlike regression, Classification produces a category rather
than a value, such as "Green or Blue," "fruit or animal," and
so on. Because the Classification method is a Supervised
learning technique, it uses labeled input data with
corresponding output, which means it includes data that has
been labeled.
In classification algorithm, a discrete output function(y) is
mapped to input variable(x).
y = f(x), where y = categorical output
Email Spam Detector is the greatest example of a machine
learning classification method.
The fundamental purpose of a classification algorithm is to
determine which category a dataset belongs to, and these
algorithms are typically used to anticipate the output for
categorical data.
The diagram below helps to understand classification methods.
There are two classes in the diagram below: class A and class
B. These classes share features in common with one another as
well as differences from other classes.
A classifier is the algorithm that performs the classification on a dataset. There are two sorts of classifications: classifications and classifications.

Binary Classifier: This type of classifier is used
when there are only two possible outputs to a classification
task.
YES or NO, MALE or FEMALE, SPAM or NOT SPAM, CAT or DOG, and so on are some examples. 
Multiclass Classifier: A Multiclass Classifier is
used when a classification task involves more than two
outcomes.
Classifications of different sorts of crops, for example, or classifications of different types of music.
Learners in Classification Problems:
There are two sorts of learners in classification problems: We

Lazy Learners: A lazy learner saves the training
dataset first and then waits for the test dataset. In the
case of the lazy learner, classification is based on the
most closely related data in the training dataset. Training
takes less time, but projections take longer.
Casebased reasoning, for example, uses the KNN method.  Eager Learners: Before receiving a test dataset, Eager Learners create a classification model based on a training dataset. Eager Learners, in contrast to Lazy Learners, spend more time learning and less time predicting. Decision Trees, Nave Bayes, and ANN are some examples.
Types of ML Classification Algorithms:
Classification Algorithms can be further divided into the Mainly two category:

Linear Models
1. Logistic Regression
2. Support Vector Machines

Nonlinear Models
1. KNearest Neighbours
2. Kernel SVM
3. Naïve Bayes
4. Decision Tree Classification
5. Random Forest Classification
Note: We will learn the above algorithms in later chapters.
Evaluating a Classification model:
It is vital to evaluate the performance of our model once it has been completed, whether it be a Classification or Regression model. So, we have the following options for testing a classification model:
1. Log Loss or CrossEntropy Loss:
 It's used to measure how well a classifier performs when the result is a probability value between 0 and 1.
 The value of log loss should be close to 0 for a decent binary Classification model.
 If the anticipated value differs from the actual value, the value of log loss rises.
 The lower the log loss, the more accurate the model is.

Crossentropy can be determined for binary classification as
follows:
?(ylog(p)+(1?y)log(1?p))
Where y= Actual output, p= predicted output.
2. Confusion Matrix:
 The confusion matrix generates a matrix/table as an output that describes the model's performance.
 The error matrix is another name for it.
 The matrix is made up of the results of the forecasts in a summary form, with a total number of correct and incorrect predictions. The matrix is shown in the table below:
Actual Positive  Actual Negative  

Predictive Positive  True Positive  False Positive 
Predictive Negative  False Negative  True Negative 
3. AUCROC curve:
 The AUC refers for Area Under the Curve, and the ROC curve is for Receiver Operating Characteristics Curve.
 It's a graph that depicts the classification model's performance at various thresholds.
 The AUCROC Curve is used to visualize the performance of the multiclass classification model.
 TPR (True Positive Rate) is plotted on the Yaxis, while FPR (False Positive Rate) is plotted on the Xaxis.
Use cases of Classification Algorithms
Algorithms for classification can be employed in a variety of situations. Here are a few examples of how Classification Algorithms are used:
 Email Spam Detection
 Speech Recognition
 Identifications of Cancer tumor cells.
 Drugs Classification
 Biometric Identification, etc.