Skip to main content

Big data analytics and gold coloured unicorns

big data analytics

Big data and advanced analytics is the new hype in the financial world. More and more financial profiles are sought with a note of these techniques in order to make the translation between the IT department and the business. This article provides an overview of the basic concepts and the different techniques.

Data analytics is the discovery of patterns and models in (enormous) amounts of data. These models must be valid, useful, unexpected and understandable. What can you do with a model that detects credit card fraud with 100% accuracy, if it takes 30 minutes per check? Is it useful then? There is also the discussion between black vs white box models. Black boxes are often very accurate and precise, but we don't know how they work. How can you defend the results in a lawsuit?

The big difference between classical analytics and big data analytics is the level of insights, ranging from descriptive to predictive and prescriptive models. Prescriptive analytics is the process of converting predictive results into business cases with real value. Currently we see three techniques that have a wide number of applications:

1. A large amount of big data applications fall into the category of classification and prediction.

Take banks as an example. Every day millions of people ask for new credit cards, loans and mortgages. In the decision-making process, banks use one number to assess someone's financial history and assess the likelihood of paying off debt: a credit score. This score is calculated based on all the information the banks know about you.

2. In the telecom sector, switching customers from one company to another is called churn.

Because attracting new customers is much more expensive than retaining new customers, companies have invested a lot of time and effort in creating and improving churn models. The aim is to highlight the customer at risk and find ways to retain them, for example through retention incentives.

3. Recommender systems are used in different domains.

They are used for book recommendations on ("Customer who bought this item also bought ..."), for music recommendations on Spotify, for movie recommendations on Netflix, and for news recommendations on almost all news portals.
The list of techniques is actually much longer and evolves every day. That is why it is more important to first define your need and then see which applications there are. As a financial business partner you can play an important role here.

And also remember this

“A data scientist is like a gold colored unicorn: mythical powers, but impossible to find”.

AUTHOR: Junior Verschuren

stay up to date