Why simulate?

At the core of an Elara Solution is a model, or digital twin, of a business or organization. Having a mathematical model of a business allows you to simulate the model on a computer. Recent advances in computing, machine learning and mathematical techniques has unlocked the ability to model the complexities and interdependencies of real-world organizations that drive business outcomes.

In the past, leaders and managers have had to rely on a few limited techniques to increase performance. Leaders can use their understanding of current performance to make big-bet, transformative changes to their organization. A manager could use their experience and "gut feeling" to guide day-to-day operational decisions. In the messy real world, where there is a complex chain of interdependcies and cause-and-effect relationships are tangled, either approach can (and sadly often do) lead to suboptimal outcomes when considering high-level business objectives.

Elara's models can simultaneously handle such a high level of detail, a broad scope of view, and simulate with such low latency that it can outperform human-made decisions where:

• there are many details to consider;
• the value chain is broad and crosses intra-organizational boundaries or 'silos';
• there are complex non-linear relationships between decisions and high level business outcomes; or
• the decision space is large and impractical to explore by hand.

In the modern world, such challenges are par for the course. Elara is able to overcome these and consider the impact of decisions throughout the value chain to recommend decisions that achieve the best performance.

The goal of this topic is to show you how to build a useful model with the EDK and produce optimized recommendations from Elara. The kinds of insights you can derive with Elara in order to maximise business objectives are discussed next.

Deriving insights

Business analytics is often described as having three distinct types: descriptive, predictive and prescriptive. Descriptive analytics is the most basic, deriving meaning from historical data so you can understand how an organization is operating. Predictive analytics attempts to predict future outcomes, allowing you to set baseline expectations of future performance and foresee upcoming challenges. Prescriptive analytics takes the ultimate step of prescribing what actions to take today to maximize the performance tomorrow. These different types of analysis build upon each other, with the most benefit derived from the top of the pyramid with prescriptive analytics.

Below you will explore how Elara guides you generate all of the above insights and optimize business performance.

Descriptive insights

Descriptive insights derive meaning from historical data. They allow you to tell a story about how a business is operating, so you can build a narrative about what is going well and what might be improved.

Traditional analytics tools can provide a limited set of insights. You might summarize data in Excel, group and aggregate statistics with SQL, or search out obvious correlations visually in PowerBI or Tableau. But correlation is not causation, and one aspect that is often lacking is any direct evidence of cause and effect.

Elara takes a different approach. With Elara you construct a model of a business and its activities. Any effects that you observe are directly caused by processes in your model.

By using a causal model you can provide much more meaningful insights about what has happened in the past. The simulation produces a series of events, each affecting the next. This allows you to build a understanding of how past decisions have affected a business. You can build a narrative that describes the cause of the outcomes you have experienced. This level of insight means leaders and managers have an accurate and precise understanding of how their business is operating today.

Perhaps the most important part of developing a descriptive model that accurately reproduces historical outcomes is that it allows you to move towards the next level on insight – prediction.

Predictive insights

Predictive analytics give you an understanding of what to expect in the future. These can be used to inform many aspects of a business, such as setting performance expectations or goals. They can also highlight any impending issues so that they can be de-risked and mitigated. The more accurate and reliable the predictions, the more benefit can be derived by leaders and managers.

With Elara, once you have a descriptive model of an operating business, it is generally straightforward to simulate that model into the future to generate predictions.

That said, producing useful and accurate predictions requires thinking about things usually not required for historical modelling. Most importantly, there is a lot of uncertainty about future events and the operating environment. Such detail needs to be predicted for a model to bve representative of the real life business. And there is a whole range of potential futures to consider.

Elara tackles these head on, by including uncertainty directly into the simulation processes. And you can train a machine learning model to learn historical behavior (based on observed features) and use this to predict forward. Together this makes creating a predictive model in the EDK a breeze.

One aspect that is captured by a model in Elara but not by more elementary methods of prediction is the chain of cause and effect. Elara doesn't simply predict the future by extrapolating trends of the past; it models the actual value chain. A decision or event can have a rippling effect through the model. In unstable situations even small changes can greatly affect future outcomes (the butterfly effect). And using the Elara platform, you can even benchmark earlier predictions against observed outcomes.

Predictive analytics provides businesses with one more powerful capability: scenario analysis. When a decision needs to be made or an idea for change is raised, two or more "what if?" scenarios can be generated and compared side-by-side. These allow leaders and managers to understand the potential benefits and trade-offs involved in the various options before committing to a certain decision.

Automating the discovery of decisions to optimize performance is the goal of prescriptive insights.

Prescriptive insights

Prescriptive insights sit at the top of the analytics pyramid. They prescribe choices to make at decision points in order to optimize future outcomes.

Elara uses predictive models together with advanced mathematical optimization in order to recommend decision choices. As an EDK developer you define the high-level business objective(s) you want to maximize. You can then ask Elara to optimize certain decisions with respect to the objective. This produces another scenario referred to as an optimized scenario, together with the recommended decisions to make and the consequent outcomes.

These recommendations can be taken on board by a business to optimize operational activities or make structural changes. Integration of business systems with the Elara platform allows the process to be automated and near-real-time from end to end.

Elara is designed to make it easy for you to produce descriptive, predictive and prescriptive analytics, by following a structured pathway.

Next Steps

In this tutorial, you discovered some of the benefits that can follow from applying a simulation-based approach to a organization. In the

next lesson you will learn how Elara uses resources to model the material, equipment, people and information core to your activities.