Data Science: A New Approach For Problem Solving and Business Strategy

Artificial Intelligence

,

Data Intelligence

Data Science: A New Approach For Problem Solving and Business Strategy

Stefano Oddone | Mar 15, 2019

 Data science, that is leveraging scientific methods, processes, algorithms and systems to automate, or at least  drive, business decisions, is a fundamental aspect of today’s business.

The new capabilities of data intelligence unleashed by the rise of cloud computing and artificial intelligence make it one of the most promising areas of the digital transformation.

Before we being to explain Data Science and how to approach the introduction of this discipline into your business, we start with aligning on the fundamentals.

Data Science - a new way to approach problem solving and business strategy

The role of Machine Learning and Artificial Intelligence in Data Science

When computing high volumes of data and trying to extract new meaning from these data, the conversation today centers on the themes of artificial intelligence (AI), with particular attention to a branch of AI coined "machine learning".

According to Wikipedia, Artificial Intelligence can be defined as the the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. The use cases for Artificial Intelligence are numerous and continue to grow exponentially thanks to constant innovations on the technical frontier.

Machine Learning refers to self-learning algorithms which mainly use statistics to draw models from huge amounts of data, where data includes: numbers, words, images, clicks, whatever. If it can be stored digitally, it can be incorporated into an automatic learning algorithm.

Applying artificial intelligence and in particular, machine learning to traditional and nontraditional Business Intelligence (think: unstructured data) can allow companies to reveal otherwise untapped trends in their business in order to create new and effective strategies lending to increased profitability, cost savings and pretty much any other goal that they aim to achieve.

As an example, refer to our blog on applications of machine learning in human resources.  

Data Science Process

Widely popularized over the last several years, there are many definitions, diagrams and methodologies to explain the process of data science. Being a visual person, I prefer to focus on a visual graphic to explain this multifaceted process:

 Data Intelligence - Data Science Deconstructed

Credits go to AJ Goldstein ( LinkedIn profile and more details available @ ajgoldstein.com )

There are a few important takeaways from this image that you should keep in mind when approaching your new Data Science team.

#1 - Don’t search for the entire set of required competences in a single person, Machine Learning and Artificial Intelligence is a team sport.

Yes, every Century we have a few of super-talents that are clearly well above the mark (e.g. Pelè and Maradona for football, Coppi and Merckx for cycling, Phelps and Lochte for swimming,...) but even those universal champions were not able to be so cross-functional to succeed in every role and specialty, (Phelps wouldn’t have won 28 Olympic medals if he had decided to play water polo). They could have won some match/race by himself alone but they needed a team, inside and/or outside the pitch, to win Championships...and we are talking about the best of the best.

#2 - Don’t approach an Advanced Analytics implementations as you would a “classic” business intelligence project; to generate new value requires new ways of thinking about and approaching the challenge.

Even if it could sounds obvious, it is worth underlining. When switching from “rule-based” analytical solutions to “data-driven” ones, the old paradigms are no longer effective, for instance the classic “business requirement gathering” phase must be replaced with a highly interactive and consultative activity of problem shaping.

At Techedge we have a methodology to successfully drive Machine Learning implementations, below a summary of key elements:

  • When it come to data, potentially huge amounts of data, a rigorous approach is key. We propose “CRISP-DM” which means a Cross-industry standard process for data mining. The methodology is, as the name indicates, cross-industry and flexible, allowing it to be applied to disparate use cases
  • On top of the methodology we add our expertise: We offer predefined templates and deliverables for each phase of the data science process to increase productivity and alignment with business users
  • If you are not totally sure Machine Learning could be a viable solution for your case, Techedge can provide a well-defined and designed assessment to help your decision makers bring clarity to the scenario, identify data availability and possible outcomes.

In the next posts on this topic, we will be addressing each of the the six steps of Data Science summarized earlier by Goldstein, going deeper into what is needed to succeed in each of them.

Subscribe for more Data Intelligence tips, news and updates!

Are you ready to start deploying a more scientific, data-driven strategy to your business operations? Subscribe to our blog to stay up to date on the latest tips, news and more!

Subscribe to our Blog

Subscribe!