Traditional vs. Modern Data Science

In the second part of our mini series about Data Science for IoT, we’re going to explore some of the many differences between modern and traditional data science. Perhaps the easiest way to tackle the subject is to look at the differences between a traditional data analytics team and a team of data scientists. The collection and analysis of customer data is nothing new to the business world. Both big and small companies use Big Data collected from various sources, such as machine logs, social media accounts and archives, to name but a few, to extract useful insight which can then be applied to their business strategies. What has changed, however, is the way in which data analysts are doing this.

A traditional data analytics team uses descriptive and exploratory analytics methods in order to reveal performance results and discover patterns. They tend to focus on current and past data and do not make predictions regarding future behavioural patterns. Data scientists, however, channel predictive and prescriptive methods of analysis to predict emerging trends. Rather than focusing on past and present data, they concentrate on what has happened, why, and what it suggests about what may happen in the future, so that the company can implement better business strategies and, therefore, see more profit.

Another significant difference which sets the two groups apart, is the use of different forms of data. For example, data analysts tend to work with structured or “clean” data, as it’s easier for them to compile, store, organise and, of course, analyse. Data scientists, however, rely on unstructured or “dirty” data for their analysis. Said data tends to come from more obscure sources, such as emails or social media engagement, and is processed by using probability and statistical algorithms to obtain insights. Such processes are beyond the knowledge of a traditional data analyst.

These are just two prime examples of the differences between a traditional data analytics team and a team of data scientists, but the list is not exhaustive.

“But, what does any of this have to with IoT?”, we hear you say.

In order to obtain valuable insight from the vast amounts of data that connected devices generate on a daily basis, data scientists must be capable of applying data science models to IoT datasets in order to extract, store and analyse said data, make real time predictions and detect anomalies. The sheer volume of IoT datasets requires traditional data science technologies to be improved and enhanced, so that we are able to discover critical insight, improve processes and make smarter decisions. So, although traditional data science paved the way for the data analytics that we know today, the analytic techniques we use for IoT are far more sophisticated and advanced, due to the nature of the data they deal with.