The Big Picture of Analytics & Machine Learning
Business intelligence, Business Analytics, Advanced Analytics, Machine Learning, Big Data Analytics, Data Mining, Data Scientist – many terminologies and varying definitions exists today, as it happens with any emerging technology and industry. A beginner will get lost, within minutes, in this quagmire of jargons. And I did too.
Until a set of standardisation is established, conflicting interests and claims will continue. The trick is not to get caught in the nomenclature, but look at the BIG picture and devise individual interpretation.
What does a business look at?
At a macro level, businesses, at any given time, look for 2 things:
- Performance: What has been done till date? Why it has happened?
- Outlook: How to shape the future?
Let’s take telco as an example. One common performance indicator telcos’ focus on is “Revenue”; every day, week, month, quarter & year, revenue is tracked and analysed. Additionally, telcos’ continuously plan to drive more revenues in the future. They devise new services, marketing strategies, and offers to fuel revenue growth.
As you observe, there is a strong bond between the past and the future. What happened in the “Past” is a vital input to devise strategies for the “Future”. Likewise, the strategies for “Future” will become key performance indicators to track continuously.
But, where to get the information
of the past and for the future?
Where to get the insights from?
Insights are spread all over the organisation in various forms. It is in documents, databases, files, reports, survey, and peoples’ head (domain expertise). To get insights into the past and the future, companies need to access this information and use them efficiently.
In many situations, the information is not readily usable. It must be converted, formatted, enriched, and cleaned. Experts handling the tasks relating to data are the “Data Engineers or Data Miners”.
The need for Technology
Data is exploding. The proliferation of smart phones, digitalisation, automation and social media are contributing to the data growth. Technological improvements like cheaper storage and RAM, faster processors and in-built analytical capabilities of IT systems, are enabling companies to gather data, that weren’t possible a few years back. Digital transformation programs, to gain insights about themselves and their customers, is the norm today for many organisations.
A wide variety of platforms are available to cater to Business Intelligence & Business Analytics. At the core is the ability to ingest the data, and provide them for insights.
| Business Intelligence | Business Analytics |
| BI platforms | Statistical Analysis tools |
| Enterprise Data Warehouse platforms | Machine Learning programs |
| Visualisation & reporting tools of IT systems | Programming Languages |
The methodology for Insights
Analytics is the route to insights. Broadly, analytics is grouped under 3 categories:
| Descriptive Analytics | Descriptive analytics is about the past, and the elementary form of analytics. “What happened in the past?” |
| Predictive Analytics | Predictive analytics is to predict future events and outcomes. “What will happen?” |
| Prescriptive Analytics | Prescriptive analytics is for optimisation & decision making. “What action to take?” |
All forms of analytics have existed for a long time. However, with the advent of technology and tools, new roles have sprouted in the recent past. Experts handling the Descriptive Analytics are typically called as “Business Analysts”, while those working on Predictive and Prescriptive Analytics are tagged as “Data Scientists”.
It must be noted that these are roles, and a fair amount of flexibility exists across companies and teams in the responsibilities. Depending on the size and complexity of problem being handled, these roles are either played by an individual or distributed across a team.
Consuming the Insights
The key to a successful analytics program is to consume the insights, and apply it in business. The output from an analytics program comes in multiple formats to consume. The essence is that, irrespective of the format, for a business to benefit from it, the output eventually needs to be applied in the context of business. Typical applications include process changes, implementing a rule in the systems or releasing a new product/service.
| Business Intelligence | Business Analytics |
| Dashboards | Forecasts |
| Reports | Scores |
| Charts & tables | Recommendations |
Traditionally, Business Analysts & Data Scientists have taken up the role of explaining the results to the business users and executives. As companies are leaning more on analytics for decision making, and businesses are finding it hard to consume technical information, a new role – “Analytics Translator” or “Light Quants” – is emerging to bridge the gap between business and technology. Like I mentioned in the previous section, it is a role and any individual with the right skills could play it.
The BIG picture
Bringing all into perspective, this is my big picture of the analytics landscape. In the context of Business Intelligence & Business Analytics, 3 elements come together –
- A business need or requirement
- Technology to enable the ecosystem of data processing and analysis
- Techniques to apply the right methodology and tools for outcomes
In the upcoming blogs, I will get elaborate on these concepts.