TotEM and CRISP-DM FAQ V6
0 likes60 views
Overbeck Analitica's Totem Total Enterprise Modelling (TOTEM) framework emphasizes the importance of developing an appropriate organizational environment for successful predictive analytics implementation. The method focuses on collaboratively deriving pertinent business questions before data modelling, which distinguishes it from traditional approaches like CRISP-DM. TOTEM has proven particularly effective in complex scenarios, such as predictive asset management in industries like water and energy, ensuring meaningful outcomes are achieved rather than just correct model predictions.
TotEM and CRISP-DM FAQ V6
- 1. FAQ: TotEM Total Enterprise Modelling Confidential. Copyright © 2015 Overbeck Analitica, All Rights Reserved. What is TotEM and where is it used? Overbeck Analitica's vision has been shaped by years of experience in analytics. As we have watched the industry grow we have seen the capability of predictive analytic components grow at a faster pace than organisations can implement and derive value from them. Software suppliers have also underestimated the organisational and cultural challenges involved in getting a predictive analytics project off the ground. In response, Overbeck Analitica developed - a road map that placesTotal Enterprise Modelling as much emphasis on creating the right organisational environment in which to use predictive analytics, as it does on the enabling software. The TotEMTM roadmap works with your organisation as a whole to create a collaborative environment across teams and departments by building consensus, connecting people, systems and data to enable the predictive enterprise. Water industry example Let’s take an example from the water industry: flood prevention. Preventing floods depends on having the right infrastructure assets (i.e. pipes), the right non-infrastructure assets (e.g. pumps) and the right operations approach to deal with the media flowing down the pipes (e.g. regular cleansing). All three need to work well to prevent floods: correctly specified and maintained pipes and pumps and regular cleansing. The trouble is that there are many, thousand miles of pipes, thousands of pumps to maintain and thousands of cleansing and unblocking activities to be carried out every year. In this example, the head of the pumps department was getting hassle because the pumps needed very frequent unblocking. He said he suspected that there was nothing wrong with the pumps, the problem was that there was insufficient money being allocated to cleansing and so
- 2. FAQ: TotEM Total Enterprise Modelling Confidential. Copyright © 2015 Overbeck Analitica, All Rights Reserved. the media was becoming worse and blocking up the pumps. This all has huge financial and emotional implications and raises numerous questions. Where do you best direct many millions of pounds to prevent flooding of people’s property and beaches? Do you put it into more cleansing or into different pumps or something else? What is the connection between cleansing activity, media quality and pump unblocking? How do you even formulate the right business question before you begin to enable the modelling of the data for these thousands of pumps and miles of pipes? This is where TotEMTM comes in. CRISP-DM is used to tackle the modelling part of this problem but it rarely challenges the business sufficiently to pinpoint the correct question to be asked. What is involved with TotEM? The TotEMTM process begins with the experienced team at OA conducting a series of structured interviews with senior execs. We know what’s really needed in order to begin a good CRISP-DM modelling cycle and that is what we work towards through a series of interviews and workshops. We then create recommendations that contain one or more carefully formed, critical business questions. Each structured question enables a CRISP-DM modelling cycle to be undertaken with questions that address very significant, previously untapped novel business value. This approach ensures that we, together with our customer stakeholders, get the right answer to the right question. When is TotEMTM not used? TotEMTM is extremely effective in situations where a complex business problem needs to be tackled by several different departments and is used a lot with sophisticated data techniques such as predictive analytics. It is not applicable where a defined set of information needs to be delivered or structured for easier navigation, so we wouldn’t use it to deliver a dashboard or develop a data warehouse for instance.
- 3. FAQ: TotEM Total Enterprise Modelling Confidential. Copyright © 2015 Overbeck Analitica, All Rights Reserved. What can go wrong when a framework like TotEM is not used? As an example of how costly modelling can be when a framework like TotEMTM is not used, consider what happened to a major international subscription media company. They had been an early innovator in a rapidly growing market which then matured. They realised they were losing customers to competitors who had entered the market later and they needed to do something about it. Being a switched-on company they wanted to use their customer behaviour data to help them decide what to do, so they formulated as their business question, “How do we retain customers?” With the aim of retaining customers, the company applied CRISP-DM and developed a propensity model to apply differential treatments to different customers depending on their different characteristics. At a practical level, they translated that into a set of different ways for the customer service team to respond to different customers who rang up to cancel their subscription. The campaign was very successful in that many customers were saved from leaving. However, despite successful customer retention, the company’s profitability declined sharply. The problem was that they were getting the right answer to the wrong question. Why TotEM is more effective than a modelling methodology such as CRISP-DM TotEMTM asks more fundamental business questions before diving into data modelling. The business issue for this company was not just to retain customers, but also to retain the right customers and to help the others leave. Some customers are not profitable, but it can be hard to spot them when you have millions of customers – hence the value of modelling. Finding the right business question. In this case the right business question derived from TotEMTM turned out to be, “What is the propensity of a customer to be saved?” This business question was identified through the business-level engagement of TotEMTM and it completely changed the subsequent modelling approach. The right answer was now being sought to the right question. The subsequent modelling using our enhanced form of CRISP-DM unearthed a significant group of customers who were repeatedly threatening to leave and being given incentives to stay. This was what made them unprofitable: they were freeloaders. The new model took this behaviour into account and identified customers whose behaviour needed to be changed or else they should be encouraged to leave. This is how TotEMTM differs from using a pure modelling approach such as CRISP-DM. Modelling frameworks ensures the right answer is achieved and TotEMTM additionally ensures it is the right answer to the right business question.
- 4. FAQ: TotEM Total Enterprise Modelling Confidential. Copyright © 2015 Overbeck Analitica, All Rights Reserved. What is the TotEM staged delivery methodology? Considering a TotEMTM approach on Predictive Asset Management as an example, our staged delivery methodology guarantees successful solution deployment. Starting with initial Proof of Concept (Vision & Scope Study) to full solution roll out (Operate & Evolve Service) enabling asset intensive organisations such as water, energy and manufacturing to shift their business processes and maintenance systems from fail-and-fix to predict-and-prevent economic maintenance regime. Predictive Asset Management … using … Reliable, fit-for-purpose data … from … Optimised business processes … delivering … Reduced combined maintenance cost and risk control preventing outages, loss and pollutions … avoiding … Business ‘Silo Effect’ constraints
- 5. FAQ: TotEM Total Enterprise Modelling Confidential. Copyright © 2015 Overbeck Analitica, All Rights Reserved. What is TotEM weekly progress/status reporting? A single point of contact, usually your appointed engagement or project manager overlooking all TotEMTM deliverables is publishing a two page summary weekly progress update report, tracking the project from kick off to successfully steering the project to agreed deliverables on time and on budget. Our TotEMTM data driven implementations are transparent and fully auditable receiving industry wide recognition.