Moving from “what happened” analytics to “what will happen”
James Beech, Senior Data Scientist BNZ/Brera Consulting Group
What’s “Big Data” all about ? This presentation will look at developments in the financial services as examples of what is coming for agriculture.
Financial services are aligning data sources to create single customer views. This requires a large overhaul of traditional systems, to build platforms to handle vast amount of disparate data (Variety, Velocity, Volume, Veracity). It offers greater understanding of customer needs so services can align the right product to the right customer using predictive modelling.
Financial services biggest change is moving from “what happened” analytics to “what will happen”. They are using new techniques to understand causes and predictions, but to do so are building teams of data scientists and data story tellers.
Geospatial data sources and government data are enabling us to get a holistic customer view. Geospatial location of the customer base gives opportunities to align to market potential. Further, the Open Data approach by local government and NZ government reflect a new view of data and information as key public assets. Government believes making data available will drive innovation through better decision making and the creation of new services, tools, and knowledge.
An example is the ANZ Truckometer. After carefully analysing the data ANZ selected key routes and applied statistical techniques to smooth out anomalies and gaps. The result is a strong correlation between traffic flows and predicting economic growth or decline as measured by GDP data from Statistics New Zealand.
Another big data example involves Cancer identification in eye images through the application of Deep learning and Machine Learning. A predictive model was trained on 80,000 labelled images and shown to predict with 87% accuracy. These algorithm parameters allow the ability to apply to other cyclical images.
An agricultural application would be an action oriented Agricultural Dashboard. To get results, developers must focus on the problem being solved, not the product. Other lessons show delivery in a timely and relevant manner is crucial, it must build solid ROI cases and that the user interfaces and user experience are very important.
Focus less on “all the data” or the “perfect algorithm”. Use a short agile framework to roll out product and services. Customers will provide you the feedback for next iteration.
Farmers need end to end solutions, not part of the problem solved.