More than 10 years ago, ERP, CRM and BI were real trendy buzzwords in business. During the last 10 years, their popularity has almost been taken over by three new buzzwords: Analytics, Big Data and Optimization. Data Science is the new kid on the block.
If we look at Google Trends for these buzzwords, Analytics now scores highest in terms of searches, both globally and in the Netherlands (see figure 1). In the Netherlands, Analytics has been more trendy than BI since 2009. Big Data is a strong runner up since 2011 and Data Science since 2014. Optimization is stable since 2007.
What are the connections between Analytics, Big Data and Optimization?
Connection 1: Analytics fuels Optimization with better insights and decision rules.
To benefit most from Optimization, especially software-based mathematical optimization with actionable results as output, companies need to work on the basic steps of Analytics first. These steps are: descriptive Analytics (also known as Reporting or Business Intelligence) and predictive Analytics (also known as Forecasting). These analytics steps are needed to learn more about data (what happened and why) and to discover optimization rules (what should happen next).
Connection 2: Big Data adds more reality and agility to Optimization
To benefit from Optimization, most companies must look into external data for more realistic Analytics results and more agile Optimization models. While most companies start with Analytics and Optimization only on Small Data (ERP, CRM, HRM data: structured, internal data), major breakthroughs are being achieved by applying Analytics and Optimization on both Small and Big Data (external data, non-structured data).
This view leads to an extension of the Analytics Maturity Curve, pictured below as the Analytics and Data Scope Matrix (figure 2).
In this matrix, two maturity challenges are combined:
- Challenge 1, related to Analytics: how to grow from descriptive to prescriptive analytics (or from reporting, via forecasting to optimization)
- Challenge 2 related to the Data Scope: how to extend the data scope from Small Data to Big Data for all analytics levels (descriptive, predictive and prescriptive)
Optimization is the goal. But what is it?
It’s very clear that Optimization, applied on both Small Data and Big Data, is the ultimate goal for your organization. However, interest in Optimization is still quite limited, according to Google Trends. The term did see a boost in popularity more than 10 years ago through books like “Competing on Analytics” and “the Optimization Edge,” but the uptake is still marginal. Why is that?
By explaining what Optimization is, we can explain the delay in interest.
A good definition for Optimization can be found in the book “Optimization for Dummies” (Saigal, 2012). Optimization is defined by Saigal as a mathematical way to represent a business process in software to recommend decisions that generate the best possible result. Optimization solutions are based on an optimization model, which include objectives, variables, business rules, constraints, a solver and relevant data.
A visual representation of Optimization can be found in the Plus Optimization Model (see figure 3). This Plus symbol was developed as part of the “Optimize Your World” program at ORTEC.
In this model, Optimization is defined as the combination of:
- a business case (with related data)
- business rules
- an optimization model
- results (based on the execution of the optimized outcome)
Analytics is used to analyze the business case, define the goals and rules, define the optimization model, analyze the results and improve the optimization model, again and again.
Small and Big Data are used to fuel analytics and add reality to all elements (case data, goals, rules, optimization model)Figure 3. Next to Reporting and Forecasting, Optimization visualized in the Plus Optimization Model (ORTEC/Buijsse, 2017)
Optimization: what are criteria for success?
Optimization leads to better results, based on a better selection of preferred options, taking into account more and better data, goals and business rules.
But to maximize your success with Optimization, you must meet many different criteria: having a clear business case, having good internal and external data, having proper, quantified targets, the appropriate business rules, constraints, optimization models etc. These criteria are often the main bottlenecks standing in the way of successful optimization implementations.
Let’s look at them through the Plus Optimization Model:
- Case data: is there enough good data to make a specific business situation “facts-based”? Are both Small Data and Big Data included, to make the case as realistic as possible?
- Goals: are goals clearly defined, measurable and prioritized, to make them fit for use in an optimization model?
- Rules: does the company fully understand what the variables, constraints and business rules are? Are rules, variables and constraints defined on both strategic, tactical and operational levels, to make the model more robust?
- Optimization Model: are the appropriate (mathematical) optimization models available to solve this type of problem? Are all criteria incorporated in the optimization model, and is the model sufficiently robust or “agile”?
- Results: is there sufficient confidence in the quality of the data, the optimization model and the outcome of the optimization processes, to make results actionable?
Analytics, Data Science and Big Data can help you meet these criteria. Therefore, increasing interest in Analytics, Data Science and Big Data will also boost the interest in Optimization and lay a foundation for its success. When Analytics, Data Science, Small Data and Big Data pieces fit neatly into a puzzle, organizations can succeed with Optimization. It is the ultimate solution for creating the best actionable results, also for real-time execution and reporting. The yearly Edelman Awards at INFORMS (http://www.informs.org/) prove the unique value of Optimization every year. In addition, the growth in optimization projects proves that more and more companies are able to successfully implement software solutions for optimization.
Analytics and Big Data: perfect triggers for Optimization opportunities?
To conclude, increasing interest in Analytics and Big Data is creating a fantastic breeding ground for applying Optimization and related opportunities, such as Data Science, Machine Learning, Artificial Intelligence and The Internet of Things. However, applying Optimization successfully requires a thorough understanding of the scope of the project (which data, which objectives, which business rules) and a good mindset. The real potential will only be enjoyed by organizations that can master challenges related to the Analytics Maturity and Data Scope, as pictured in the Analytics and Data Scope Matrix. To learn more about Big Data and Analytics or read Optimization success stories in Supply Chain and Workforce Planning, visit http://www.ortec.com/.
For an explanation of all acronyms and terms in this article, see http://data-informed.com/glossary-of-big-data-terms/
Many thanks to ORTEC colleagues Adriaan Tas, Ronald Buitenhek, Gerrit Timmer, Bart Veltman, Sicco Brakema and Lex Knape for their input to this blog.