Machine Learning — Not rocket science!🚀

Manasi Anantpurkar
Fields Data
Published in
5 min readOct 14, 2021


A robot looking at a tablet-like device, giving an impression of a machine trying to learn
Photo by Andrea De Santis on Unsplash

Are you surprised when Netflix suggests shows similar to those you watched earlier? Do you ever wonder how the promotional emails you receive automatically get sent to your spam folder? How do Google Maps know there is traffic on your daily route and suggest a better alternative? The technology behind all these wonders is Machine Learning.

In this blog, I am going to explain Machine Learning in the simplest terms so that you’ll know what it is about, next time it comes up in a conversation. As our organization mainly works in the humanitarian sector, I will also focus on how similar organizations can use Machine Learning. Happy reading!

What is Machine Learning?

To understand what Machine Learning is, let’s imagine that it’s the weekend and Emma’s friends have invited her for lunch. She is wondering whether or not she should go. While making this decision, she considers several factors including the cost of the afternoon and how far the restaurant is from her place. When faced with similar invitations in the past, this is what she decided:

The green dots represent the times when Emma went out for lunch, and the pink dots represent the times she decided not to go.

As you can see, Emma went out most of the time when the cost and distance were lower, but stayed home when the cost and distance were higher.

Now, if the yellow point in the graph represents the cost and distance for Emma’s latest lunch invitation, can you guess whether she will go?

Yes, you guessed it right! Chances are she will not go for lunch.

But what if the point laid somewhere in between, as shown below?

Now it is more difficult to predict, but try drawing a circle around that point and guess again.

As you can see, there are more green dots in the yellow circle, so it’s more likely that Emma will decide to go out.

This is exactly how Machine Learning works: it predicts future actions based on previous patterns of behaviour. This Machine Learning algorithm is named “K Means Nearest Neighbours”.

We train the computer by providing it with plenty of data drawn from previous experiences. The machine then uses this to learn and understand the patterns and relationships within this data. In our example, we provided the computer with data relating to ten instances in which Emma made a decision on whether to go out, based on costs and distances. Using this interpretation, the machine can predict some outputs on random test cases. For the eleventh case, it will use Emma’s previous behaviour to predict her next decision.

The process of giving the machine data relating to our decisions in different situations is called “training the model”. Once the model is trained, we give it different situations which the machine is supposed to use to predict new decisions. This phase is called the testing phase.

Machine learning for us

Fields Data collects data about organisations in various countries working in different sectors. This data enables us to analyse numbers according to sectors, find relationships between different organisations and consolidate findings to help organisations collaboratively scale their projects.

One way in which we use Machine Learning is by creating circles in which we group similarities, each circle representing a sector for instance.

As an example, if we have the following data:

After applying a clustering algorithm in Machine Learning, the final data is as shown:

This enables us to easily identify which organisations are working in the same sector, or which organisations have overlapping projects. Similarly, Machine Learning can be used in many ways to analyse data.

Benefits of using Machine Learning in the humanitarian sector

Using Machine Learning can be especially helpful for an organisation’s maintenance, customer analysis, and security.


Data-driven organisations like ours process large amounts of data. While doing so, there is a manual requirement of entering data into the system. Machine Learning can help to clean the data by preventing duplicate and inaccurate entries. It can also help in taking preventive and correct measures to maintain company systems. For example, Machine Learning can warn you that the machine shut down last time you did the procedure that you are about to do. It can also maintain the data by detecting spam and incorrect information in the system.

Customer Analysis

Machine Learning is used to predict users’ behaviours. It can help to uncover business insights and is useful for user segmentation that enables you to create experiences tailored to your client's interests. Through Machine Learning, you can easily predict how long a customer is likely to stay with a company or on your website, and what might be their interests and further actions. More advanced uses can generate user-specific recommendations, as is seen on online shopping sites these days. While collecting data from a particular country, you can also use their local language in the system as a customised experience to make it easier for users to navigate your website.


Security is one of the most important factors in handling big data. Machine Learning can help you detect threats or unusual activity in the organisation’s system, and create warnings to draw your attention to them.

Machine learning has gained popularity in the past few years and is one of the most important, exciting, and reliable fields in the future. In the next edition of the blog related to ML, we will discuss different Machine Learning algorithms and how we can apply each one to support organisations.

So next time Google suggests you a better route, don’t be surprised and enjoy the wonders of Machine Learning!