Understanding your customers through clustering – Explained in layman’s terms

Understanding your customers through clustering – Explained in layman’s terms

Hello Datanerds,

Hope everyone is doing good.

As we start a fresh week, let’s make it productive by learning something new everyday. In this post, I am looking forward to explain unsupervised learning techniques in simple terms using a normal, day-to-day example. Hope you enjoy reading my article and if you have anything to take back any new insight from here it would make my day count! So let’s begin.

For everyone who is wondering what the term “unsupervised learning” actually means, it is nothing but a type of analysis where you are looking to get an overall view of what the general data looks like without having the intent of drawing any particular insights.

Consider yourself as a restaurant owner who is just starting to set up your own place. Once you have your place finalized, your next concern is to what to put in your menu to bring in customers to your hotel. In that case, the problem at hand in much broader in scale and you are only trying to get familiarized with your customer base. Since there is no set start point to begin to analyse the data, here is where unsupervised learning algorithms come in useful.

The most prevalent type of unsupervised learning is the clustering mechanism. As the name indicates, a larger population is grouped into clusters categorized based on similar set of characteristics. In our case, one way of grouping our population could be based on age, health preferences, calorie consciousness etc.This type of grouping helps us to get a high-level overview of what the general population/customer preferences look like and helps us to get a rough idea of what to put on the menu to attract a diverse population. Further,the clustering algorithms themselves are of many types such as Hierarchical,K-means etc. The choice of clustering model usually depends upon the data and task at hand.

On a last note, clustering techniques are really helpful in initial stages of population analysis and provides us with an overview of what our data is composed of. It has a lot of real time application such  as marketing segmentation, sentiment analysis, social media marketing to name a few. Next time you are fazed by a ambiguous statement at hand, do check out applying clustering analysis to get a better sense of the problem you’re dealing with.

Enjoy your week! 🙂

Happy Learning!

Krithika Shankar

 

 

 

 

Leave a Reply

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