Let’s talk machine learning


What is machine learning. Well in the simplest terms its training a machine. It is a method of data analysis that automates model building. It is based on the idea that systems can learn from data, pick out patterns and make decisions without human interference. It is not new science, but it has gained momentum in the recent year largely due to an increase in the amount of data generated and increased data storage options and increased computational powers. 

Why is Machine Learning Important? 

Machines learn and provide intelligent insights through sophisticated use of learning algorithms. In Order to provide value to the business, it is trained to pick patterns from data and then autonomously make decisions on new and changing data. This initiates a continuous feedback loop that generated more models to get better insights with less human inputs. 

Machine learning is used to automate tasks. It makes suggestions and predictions based on analyzing the data and performs tasks that require human intelligence i.e image recognition and speech recognition. In real life, friends’ recommendation on Facebook and Jumia’s recommendations is machine learning in its simplest form.

Types Of Machine Learning

Supervised Learning

This type of learning has algorithms trained using labelled examples. I.e a target or outcome variable which is to be predicted from a variable set of predictors. The learning algorithms are trained to generate a function that will bring the desired output. The learning algorithms learn by comparing its actual outputs with the correct output until the model achieves the desired level of accuracy. The types of supervised learning include Random Forest, Regression, Decision Tree KNS and Logistics regression.

Unsupervised Learning

In this learning, the algorithm has no desired target or outcome variable. The machine has to learn what is wanted by picking patterns from the data provided. It is common in transaction data i.e segments of customers for target marketing. Types of Unsupervised learning include the Apriori algorithm and K means.

Reinforcement Learning

This learning has algorithms that are trained to make specific decisions. It is mostly used for robotics and gaming. The algorithm learns through trial and error learning from past experiences to make the best decisions. The algorithms widely used is Markov Decision Process.

In conclusion, the increase in usage of machine learning in many industries will act as a catalyst to push data science to increase relevance. Machine learning is only as good as the data it is given and the ability of algorithms to consume it. Going forward, basic levels of machine learning will become a standard requirement for data scientist

Leave a Reply

Your email address will not be published. Required fields are marked *