###### This column is authored by content analyst and tech writer, Victoria Ashley

With the fast mechanization brought about by the technological revolution, the word manual is slowly getting lost amidst the crowd and will very soon completely vanish. As Big Data has whisked the tech industry, Machine Learning is gaining importance and has robustly handled huge amount of data making accurate predictions.

In an era of constant progress, we can only guess what astounding invention and discovery is to come next. The data-crunching machines that have been seamlessly executing the advanced techniques.

#### WHAT IS MACHINE LEARNING

Machine learning is a subset of the Artificial Intelligence, which is a broader term and concept. Where Artificial Intelligence aims to make computers smarter and intelligent, Machine Learning has come up with ways to do that. In short, it is application of Artificial Intelligence. With the use of algorithms, that iteratively learn from data, machine learning improves the functionality of computers without being explicitly programmed.

#### CATEGORIZATION OF MACHINE LEARNING ALGORITHMS

If you are a Data Scientist or a machine learning passionate, you can work your way around machine learning projects using categories in which the algorithms of machine learning have been broken down.

**SUPERVISED LEARNING**

Using a pre-defined set of “training examples”, the program is trained facilitating its ability to reach on conclusions when new data is fed. The program is trained until the desired level of accuracy is reached.

**UNSUPERVISED LEARNING**

The program is given a bunch of data and it has to detect pattern and relationship on its own. The system has to infer function to describe the pattern from the unclassified data.

**REINFORCEMENT LEARNING**

Here the system interacts with the environment and produces actions discovering errors or rewards.

All three techniques are employed in the ten common machine learning algorithms.

### THE TEN ALGORITHMS

#### 1. LINEAR REGRESSION

The algorithm can be visualized by placing items according to their weight. But the problem arises when you can’t actually weigh it, you have to do a bit of guessing here by looking at its height and width. With this visual analysis, you reach on a result. This is how linear regression works.

A relationship is formed by mapping the dependent and independent variable on a line. The line is called regression line and is represented by Y= a*X + b.

Where,

Y= Dependent Variable

A= Slope

X= Independent Variable

B= Intercept

The coefficients a & b are derived by reducing the sum of the squared difference of distance between data points and the regression line.

#### 2. LOGISTIC REGRESSION

From a set of independent variables, discrete values are estimated. It helps in the prediction of an event by fitting data to a logit function.

The following methods are used to improvise the logistic regression models:

- Adding Interaction term
- Eliminating features
- Regularizing techniques
- Using a non-linear model

#### 3. DECISION TREE

It uses supervised learning algorithm to classify problems. Decision tree is the support tool that uses a tree-like graph for decisions or probable consequences, chance-event outcomes, resources costs and utilities. The population is split into two or more homogeneous sets based on the independent variable.

#### 4. SUPPORT VECTOR MACHINE

In Support Vector Machine or SVM, raw data is plot in n-dimensional space. The value of each coordinate is then matched to a particular coordinate, making it easy to classify data. Lines called classifiers are used to split the data and plot them on a graph.

#### 5. NAIVE BAYES

Naive Bayes works on the assumption that every feature is independent of another feature. Even if there is any relation it considers each individually when calculating the probability of an outcome.

It is not only easy to use but it uses massive data sets efficiently. It outperforms even the highly complicated classification systems.

#### 6. KNN (K- NEAREST NEIGHBOR)

This algorithm is applicable to both classification and regression problems. Within the Data Science industry, it’s more often used to solve classification problems. The simple algorithm is capable of storing all available cases and classifying any new cases by taking a majority vote of its k neighbors. The case is then allotted to the class with which it matches the most. A distance function performs this measurement procedure.

Things to bear in mind before considering KNN:

- It is computationally expensive
- Variable should be normalized
- Data requires pre-processing

#### 7. K-MEANS

This unsupervised algorithm is used for solving clustering problems. The Data sets are listed into a particular number of clusters in such a way that within a cluster all data points are homogenous, and heterogeneous from the data in other clusters.

How clusters are formed:

- The algorithm picks points called centroids for each cluster.
- The data forms clusters within the closest centroid.
- New centroids are created based on the existing cluster members
- Distance between each data point is determined. The process is repeated until the centroids do not change.

#### 8. RANDOM FOREST

Together the decision trees are called Random Forest. To sort a new object based on its qualities, each tree is sorted and classified, and then the trees vote for a particular class, those with the most votes is chosen by the forest.

This is how each tree is planted and grown:

- If there are N number of training set in the cases, then a random selection of N cases is taken.
- The input variables are M
- The trees are grown to the largest level with no cutting and pruning.

#### 9. DIMENSIONALITY REDUCTION ALGORITHMS

With the vast amount of data being stored and analyzed, it is challenging to identify the multiple patterns and variables. Dimensionality reduction algorithm such as Decision Tree, Factor Analysis, Missing value ratio, and Random Forest helps in finding relevant data.

#### 10. GRADIENT BOOSTING AND ADABOOST

These are the boosting algorithms which are used when handling massive amounts of data to make accurate and speedy predictions. Boosting is a combination learning algorithm that combines the predictive power of several base estimators to improve effectiveness and power.

To sum up, it combines all the weak or average predictors to build one strong predictor.

### VERDICT

We’ve covered only basic theory surrounding the field of Machine Learning here, but of course, we have only barely rubbed off the surface.

To really apply the theories contained in this introduction to real life machine learning examples, a much deeper understanding of the topics discussed herein is required to understand its intricacies. There are many subtleties and pitfalls in Machine Learning, it can be called labyrinth where it is easy to lose your path. When it just appears to be a perfectly well-tuned thinking machine, it’s not. Almost every part of the basic theory can be tested with and altered endlessly, and the results are often interesting. Many branch into whole new fields of study that are better suited to particular problems.

The tech industry is flourishing rapidly each day, if you possess an ardor for machine learning it is worth consideration. You will find it every bit interesting and your career too will boost at a quickening pace.

Nicely done! Thank you!!