In my previous post, I have tried to explain the evolutionary process of Data Science. Since it is a broad discipline, in regards to the context of the defined problem, Data Science may include several fields, from statistics and machine learning, to marketing and process science.
In general, Data Science is a multidisciplinary field which includes combining business with statistics and software development. What does that mean? In the business world, Data Science is all about data and turning it into a real value. The possibilities are endless. Using predictive analytics for what-if scenarios and decision-making purposes has grown to be the essence of strategic planning. On the other side, advanced analytics is crucial for the optimization of operations pillars within the organization. But how is all that being realized?
Well, the first step in Data Science is to define a business problem that needs to be solved. For example, a business problem may be defined through some KPIs measured. Like low conversion rate, increasing churn rate, or high operational costs. There are methodologies explaining what does it mean to define a business problem, but in short – besides the identification of the problem itself, it includes broader definition – in terms of all the data sources and KPIs that describe the problem and should be used in the analysis. It is also important to answer the main questions – why that particular problem should be solved, and what are the main benefits? One of the most frequent mistakes managers do is that they catch a buzzword (that in most cases they don’t fully understand) and believe that it is a magical wand that could solve every problem. What is important is to prevent the “l’art pour l’art” kind of objectification.
The solution of the business problem lays within data. It is often said – data tells a story. One should just dig deep enough, and the answers will be revealed. However, in reality – that is not always the case. Data only reflects history. Which in most cases is driven by some previous actions and reactions, defining a given set of circumstances, by which a defined problem is observed. Statistics is there to help us ensure we have a representative dataset and understand the main dependencies among the KPIs. By using statistical analysis, we can determine the predictive power of the dataset and understand its constraints. Furthermore, statistics is the heart of machine learning and a crucial part of advanced analytics, through which we can analyze patterns within the data, model the historical behaviour, and get projections on the future.
Software development plays an important role in Data Science. Managers often want solutions that are reliable, scalable and automated – which requires incorporating software development into the analytics process. That means that all the algorithms should be placed within scripts, wrapped in fancy containers and callable though some API services. This is where one of the biggest problem of applying Data Science in business shows up – managers think that Data Science can be handled by one person only, while it should be the a task for a whole team, including individuals with different set of complementary skills, from business and soft skills, through programming and IT, to math and statistics. Another issue is that a lot of organizations suffer from a lack of data literacy, no defined data strategy and assessment, and poor (or no) experience with the business digitalization and transformation. These are the main prerequisites of applying Data Science in business – successfully.
When done right, the applications of advanced analytics in business context are numerous. From predictive analytics, to what-if scenarios and process mining and optimization. From retail, banking, insurance, to oil and gas, aviation, telecommunications,… Everything you can think of – can be a fertile ground for advanced analytics. The idea is simple – the data tells a story. A story that may give the answers to the questions like:
- how to increase customer satisfaction/improve customer experience?
- how to maximize ROI (return of investment)?
- how to optimize CAPEX (capital expenditures?)
- what can be done in order to increase performances?
- why is the churn rate so high?
This list is only a short preview of questions managers often ask. However, the advanced analytics and the existing methods are not almighty, and not everything can be answered by looking into some data samples, but one thing is for sure – “all models are wrong, some are useful”, and that is what it is all about, in this infinite game of enduring in the business – to search for the useful models and apply them in order to leverage the business.
In the next blog post, we will talk about Machine Learning and its applications, as well as its connection to Data Science and Artificial Intelligence. Stay tuned!