Back to Overview13 lessons learned during this most unusual summer
Ok, it probably won’t make the list of best summers in years and I have definitely been missing the carefree travelling and music festivals, but it was beyond doubt an interesting summer!
Time series data are everywhere, that is what we have been saying for quite a while now. However, as a product company our focus has mainly been to further develop and improve our product, TIM. We weren’t actively looking for potential use cases in industries we have little experience in. Eventually, our partners were the ones who saw the value TIM could add in their industry and for their customers. It is their expertise, combined with TIM’s analytical capabilities that results in an interesting value proposition.
Learn about getting more business value out of Machine Learning and Predictive Analytics
Following one of the biggest hypes of 2020, virtual events, we seized the opportunity and hosted a series of webinars together with our partners. This turned out to be an eye-opener to all of us, and a confirmation that the partner approach we kicked off early this year truly is the way forward for Tangent Works.
Our partners are taking TIM on a trip around the world to incredible places and fascinating use cases. A lot of insights emerged from their efforts, so I decided to summarize the 13 lessons I learned this summer during our Brain Snack webinar series:
- Today, many companies are using static forecasting models. When the COVID-19 pandemic hit their market, these companies were left in the dark. TIM can easily adapt to new realities, since it offers dynamic models and forecasts. Structural changes such as the one following this pandemic should no longer cause companies to be completely blindsided.
- “Inventory is in fact nothing more but a buffer for variability. Lowering variance allows to maintain a high service level whilst keeping less inventory.” – James Smith, Demand Data
- Monitoring demand variability can be achieved with the right data. By using the available internal and external influencers like costs, market trends, weather, events and promotions you can model the known variability. Next, a tool like TIM can also help you to easily experiment to model the unknown variability.
- Understanding why a forecast is the way it is greatly improves acceptance by maintenance and operations teams. Explainable AI as offered by TIM allows to continuously improve production as it gives you a better understanding of your machine and process, including interactions that are too complicated for the human brain to see.
- “AI fuels the digital transformation with big benefits. Machine Learning has matured, the technology is ready and allows scalable deployment at low cost. Keep your eyes on the cost reduction made possible by AI and avoid getting stuck in experimenting and piloting.” – Philippe Thys, ACONDA (BE)
- Machine settings, upgrades and operations change over time and each time you have to rebuilt and retrain your models or your predictive maintenance thresholds will fail to warn you. TIM can immediately adapt to new situations and retrain the models. This way, the machine can be followed continuously, even when it enters into new operational conditions, and anomalies can be detected at an early stage.
- The utilities environment is becoming more and more complicated with lots of data streams. New trends like renewable energy, district heating and smart grids are resulting in a data flood. Portfolios constantly change, meaning that data changes. The market is dynamic and so is the data. How do you cope with this and what data can help to do so? Only a flexible augmented analytics tool like TIM can provide the answers, by offering near real time model generation and model application.
- “Because gas and electricity became commodities, it is now all about the correct price, managing churn and lowering service costs while building brand awareness, increasing customer loyalty and improving the service level.” – Dirk Michiels, Tangent Works
- Since the start of the anti-coronavirus measures, electricity consumption in Belgium is 15% lower than expected. The existing, static models did not see that coming and greatly overestimated the grid load. Simultaneously, the increased share of renewables in the production mix and the lower consumption were driving down energy prices. Adaptive forecasting models such as those built by TIM can take these new market conditions into account more quickly and in turn help to avoid unexpected price swings and grid imbalances.
- The banking industry generates a lot of data and there are already many AI success stories to be found, varying from predicting mortgage volumes, to detecting fraud and forecasting customer payments.
- “Once you have your data a bit prepared you can start using TIM and get immediate results. You are limited only by your imagination as far as how you want to set up and conduct your experiments and create your predictive analytics” – Greg Oliven, Polygon Research
- TIM adapts to the timescale you need, whether that is seconds, days, months or something else. This automated flexibility is crucial in the financial industry and will help to reduce the current costs and complexity of data science as it allows for the same approach/tool to be applied across a large variety of use cases.
Did I say 13 lessons learned? Well, besides this industry-specific knowledge, our efforts in integrating TIM with BI and data analytics platforms like PowerBI, Qlik and Alteryx are also starting to bear fruit.
While the great accuracy of our forecasts gives the users of these platforms better insights to base their decisions upon, it are our amazing speed and real-time capabilities that are truly revolutionizing the industry. A BI dashboard can of course only fully live up to its potential with live data and we don’t just provide the data, but the forecasts. By automating the data science we move straight to the insights.
Rewatch our Brain Snack webinars and get to know our great partners and the fascinating use cases in which they are using TIM!
About the author
Sam Verdonck is partner manager for Tangent Works. He has a 7+ year track record in international business development, project and partner management in the fields of industrial maintenance, reliability and process optimization focusing on innovative solutions in condition monitoring, IoT and predictive analytics. Sam strongly believes in partnerships as the fuel for mutual growth.
Prior to Tangent Works, Sam managed partners and projects for ConMEC and Performance for Assets in the Middle East, Europe and South America.
