by Ea and Jan Kees
Fuelling Data-driven Insights
It is often said that ‘data is the new oil’. The process of turning data into fuel, however, is full of obstacles and seldom easy.
In this blog post we illustrate how we support our Field Partners in this process. They’ve got the boots on the ground to ensure safety and security in nature conservation areas. We’ve got the tools and engineers to create the models that can estimate risks and suggest patrol routes.
By working closely together, we fuel an inspirational learning cycle in which insights from the rangers are being used to optimize our models. And in which our models are being used by rangers to significantly increase their efficiency and effectiveness.
The result: rangers showing up at the right place at the right time with the right team to stop poaching.
How it works
The process of turning data into actionable information
Patrol > Collect Data
The first and foremost important step is to start recording the patrols of the rangers. These patrols are seldom random. On the contrary: they are based on intimate knowledge of the conservation area and the multitude of threats that the rangers are confronted with on a daily basis. Recording these with the Cluey Data Collector and Tracking app allows rangers to learn from their combined experiences.
Construction of Risk Models
Many people confuse hotspots with risks. Indeed, a hotspot is a location where many observations have been made. But that may simply be explained by the fact that rangers have often inspected that location. It does not mean that other places cannot be just as good (if you are looking for rare species) or bad (if you’re cleaning up snares).
By combining the observations made with a value that expresses the actual presence of the rangers, insight is gained in the real distribution of risks. Moreover, black spots that have not been visited at all become apparent.
Hotspot map (or heat map)
Likelihood map (or risk map)
Note that at the hotspot map center spaces glow up while on the likelihood map the edges glow up. The information conveyed in the risk map points into the direction where more of its kind (good or bad) can be found, even if these places have not yet been patrolled (in this case: the red circle).
Quarterly and Daily Patrol Advise
As risks may vary from season to season, the clever thing to do is to compare the next few months not only with the risks of the past few months, but also with the same period in preceding years. While the implicit knowledge and experience of the rangers remain very important, their encoded shared experiences can now be used to sharpen them. People always have their personal biases. By using the observations logged by all rangers, this bias is minimized. By using predictive and prescriptive models we break away from rusty patterns and gain new insights.
While quarterly patrol-advice is useful to divide the conservation area into manageable portions, daily advice should be treated as suggestions. Indeed, each day is full of surprises. One day there is a bushfire. The next day your vehicle gets stuck in the mud. And the next you come across three poachers who keep you occupied for the rest of the day.
This does not mean, however, that patrol routing advice is useless. By using Game Theory, routes are optimized to visit as many risky areas as possible, while at the same time randomness and unpredictability are built in to avoid poachers from finding a way to evade the patrols.
Quick Response
living roadmap
One of the very handy side-products that we created in the process of optimising patrols is the living roadmap. It is created from tracks that rangers actually covered and consequently always up-to-date and highly accurate. In places where roads may be flushed away during the rainy season, overgrown by lush vegetation in months, or temporarily created or cleared by bulldozers or scapers, accurate roadmaps are hard to get by.
The value of these living roadmaps is multifold.
First, it is used by the patrol advice algorithms to suggest routes.
Second, it can be used by area managers to plan park logistics and maintenance works.
Third, it can be used to quickly calculate the fastest route between you and your friend who just reported that he is in urgent need for help.
And fourth, it can be used to choose positions for setting up pop-up checkpoints or lay ambushes to intercept poachers that are fleeing the crime scene.
Functional Analytics - ownR®
The above is an example of what rightfully can be called: Functional Analytics. It is in fact the name of one of our valued Solution Partners without whom the development of the above applications would have been much harder and much more costly.
Functional Analytics is the supplier of ownR® . This platform allows analysts to work with known languages such as R, Python and Jupyter Notebooks to create any type of report, statistic, algorithm, or even neural network:
natively on your own data,
secure within the confines of your own organisation.
Moreover, and this is where it becomes even more exciting, the results can be served to end-users in the form of simple apps.
The ownR-platform has been fully integrated within the Wildlife Insights tool suite. The components of ownR that are destined for high-end users such as analysts and engineers are integrated into wildCAT (an acronym for Wildlife Conservation Analyst Toolbox). The resulting apps destined for end-users such as managers and rangers are called wildApps. All of the above apps are examples of wildApps and available to all our Field Partners.
Wildlife Insights tool suite: making things as easy as possible, but not easier than that.
The ownR-platform of Functional Analytics thus allows us to bring highly advanced analytics and data science tools within reach of nature conservation organisations. These organisations may or may not have the skills to develop wildApps themselves, but in many cases will no longer have to. Every wildApp that is being created can be used as-is, or adapted by our Field Partners.
In short:
High-end developers can skip the costly process of data engineering and work directly on their own unified data.
Managers and rangers can reap its fruit by simply clicking a few buttons.
get started in a day
Setting up your project and getting started usually doesn’t take more then a day. Maybe two. On our offerings page you can read about the various services that we supply to our Field Partners and the related remunerations.
Sensing Clues Foundation is a registered non-profit volunteer-driven organisation. As most of our work is sponsored by our Solution Partners and carried out by our extensive network of volunteers, the total cost of ownership for advanced functions like the above cannot be beaten.
About the authors
Ea Werner, born and raised in Amsterdam, is currently writing her Masters’ thesis Computational Science on this project. She has a background in Econometrics and Operations Research and a passion for anything related to the conservation of our beautiful planet!
Jan Kees Schakel founded Sensing Clues. He holds a PhD in fast-response organising, an MSc in Natural Resources Management, and an MSc in Business Information Systems. Next to leading Sensing Clues, he is strategic advisor for the Special Operations Division of the Dutch Police.