An analytical liaison for a sustainable future
Dependent on the unpredictable? To disclose the secret of upcoming environmental and weather conditions is a fundamental challenge for the energy, utilities and agriculture industries. Reliable short-term and long-term predictions of plant growth, yield, pest infestation, wind speed, rainfall, ocean currents or solar radiation – the answer lies in the analytical liaison of geotemporal information and the Internet of Things (IoT). Or in short: Data to models – this it is what the latest developments from the Microsoft Research labs are all about.
Curious about how this will reshape business models, risk management, forecasting and security of supply? Then meet the Microsoft Research team at Hannover Messe (HMI) from April 13-17 in hall 7, booth C48 in Germany. At HMI, explore three areas of our vision in connected devices, geotemporal information and cloud analytics for business development: FetchClimate, predictive modeling, and NodeAtlas.
FetchClimate: accelerate access to spatiotemporal information
FetchClimate is a fast and flexible spatiotemporal information retrieval cloud service that can be accessed via a web browser. This current public service contains a variety of common climatological information layers (e.g. temperature, precipitation) and the underlying software has been released to allow others to build their own FetchClimate. FetchClimate was inspired by the research teams’ frustrations in finding and processing environmental data. Now they’re living the dream of simply being able to specify what they want for when and where in confidence that they’ll receive the best data suited to their purposes, fast!
NodeAtlas: discover and add spatiotemporal information through a browser
Node Atlas is a cloud SQL database service coupled with an attractive user interface. This currently unreleased prototype takes working with any data that can be tagged to geographical space and time to another level. Node Atlas allows users to simply discover any information pertaining to space, time or any other number of databased properties (e.g. looking for all wind farms in Europe). New information sources (such as a new sensor network deployed in the field) can be added in real time making NodeAtlas well suited to this Internet of Things age. Advanced features like iterative search and through-the-browser schema modification not only allow users to zero in on what they want faster but also allows database developers to quickly customise existing databases to user requirements.
Machine learning meets domain understanding meets cloud information and analytics: our modelling environment
Many big data analytics applications involve combining spatiotemporal information and machine learning techniques in order to produce useful predictive models. Our new modelling environment enables that to be done rapidly. We will show how we use our modelling environment to train a new predictive model to data fetched from FetchClimate, propagate uncertainty to produce probabilistic forecasts, visualise outputs, easily iterate on the modelling pipeline to improve predictive ability and then push the model out to FetchClimate as a service. Producing valuable new environmental information has never been so easy!
The Future is Now!
