At a recent gathering in Houston, I spoke with a guy who told me his company is looking for a data unicorn. The term encapsulates what an ideal data scientist is: a coder, a statistician, a communicator, a fundraiser, a problem solver and more rolled into one.
Some have said a data unicorn is a rare find. I say it’s impossible. There’s no one individual who can be all of what a so-called data unicorn is.
The Cross-Disciplinary Reality of Data Science
My colleagues and peers consider me as a data scientist. I consider myself, though, as an agronomist with knowledge of data mining and artificial intelligence.
Yes, I can write codes and analyze data. But I rely on an in-house team of professionals as well as experts from the organizations we partner with to correctly interpret data and come up with solutions to real-world problems.
This is especially true for agriculture. Each country, and even regions within a country, has different conditions, which run a gamut — climate; farming practices; soil fertility, moisture and temperature; and socio-economic circumstances, among others.
As such, we need to know the context of the place to interpret our data properly. This means collaborating with technicians of local farmers’ associations.
A Real-World Example
Recently, we studied data on rice farming in Nicaragua as part of a project we have there. Our team ran the analysis and showed that farms which had used more nitrogen fertilizers had lower yields. One would think immediately that based on that, the recommendation would be to stop using nitrogen fertilizers.
As an agronomist, I knew that it didn’t make sense. Fertilizers should increase not decrease yields. So we dug deeper. We asked experts within CIAT and our partner organizations to know what was really happening.
From there, we discovered that nitrogen attracts a specific insect. And this insect carries a virus that destroys rice crops in Nicaragua. After that, we sent our findings to our partner in Nicaragua.
If I want to know how climate might affect farmers over the next few months, I’d go to another team at CIAT that does seasonal weather forecasting. We normally combine the forecast with information that helps farmers make better decisions, such as when exactly and which varieties to plant during a particular season.
We then share this forecast with farmers’ associations, which can disseminate the information to their members.
The Data Science Collaboration
That’s what data science is about – a collaboration among experts from various areas. Without that collaboration, I and several colleagues wouldn’t have the honor of being twice recognized by the United Nations for using big data (2014) and information and communications technologies to address climate change (2017).
Data unicorns are a myth – plain and simple. What any organization needs to do is to build a team of experts whose skills complement each other and work with external specialists if needed so they can make magic to develop real-world solutions.
By Daniel Jiménez and originally published as Everyone’s looking for a data unicorn. There’s no such thing.
Thank you for this excellent article. I would be very interested in knowing more about how you combine climate forecasting data with information that helps farmers know what to plant. Is there a systematic process for doing this? What are the most significant factors you look at, along with climate, in advising farmers what to plant during a given season? Any thoughts you have on this would be useful in helping us find ways to build farmer resilience to changes in climate.