(Header: Daily harvest from a rural village in Bihar, India. Credit: Alisa Zomer.)

The connectivity and interdependence of global agriculture makes for a complex market. Understanding how low rainfall will impact local farmers in Tanzania, maize markets in Kenya, and investment decisions at Morgan Stanley in New York City presents a dynamic political economy puzzle and a data scientist’s dream (or nightmare).

Forecasting global agriculture, like predicting the weather, involves complex modelling combined with the “gut feeling” of market experts. Unfortunately, the data are often hard to find and unwieldy to use, and predictions about things like crop yield and price are often made based on ad hoc intuition. This is where Gro comes in.

Gro, a platform created by Gro Intelligence, collects, cleans, and standardizes difficult to use agricultural data to create a one-stop information shop for decision-makers in the agricultural industry. Available on their website and as an API, the data are now easy to find and easy to use. Gro also provides their own predictive models that replace ad hoc intuition with a more consistent machine learning process. Gro Intelligence’s founder and CEO, Sara Meker, joined our seminar series to discuss their approach to agricultural data science and efforts to combat food shortages.

Presentation slides demonstrating how Gro’s forecasts compares with USDA’s forecasts and trade estimates. Credit: Sara Menker.

Improved forecasting to increase access to capital

Gro Intelligence aims to help different actors in agricultural markets, like commodities traders, supply chain professionals, and lenders, make informed, data-driven choices. Gro’s analysis provides insight into a number of factors affecting agricultural products on the market, as well as predictions about the volatility along the way. According to Sara, their yield forecast analytics are both more frequent and more accurate than the leading predictions from the United States Department of Agriculture’s World Agricultural Supply and Demand Estimates (USDA WASDE).

A former Morgan Stanley Vice President in Commodities, Sara Menker founded Gro Intelligence to address her growing awareness of the information gap in the capital infrastructure around farmers – and the resulting inefficiencies. By improving agricultural forecasting, Gro Intelligence’s intent is to assure lending markets and consequently increase farmers’ access to capital. As Gro expands into India and other countries with less developed agricultural technology (“agtech”) and markets, Sara hopes that the insight they bring will make life easier for farmers (i.e., by increasing access to capital).

Beyond drought detection to prevention

This same insight, Gro Intelligence argues, can also be used to detect famine faster and more efficiently than we’re doing right now. Famine, or extreme food scarcity, can occur as a result of droughts but also involves complex economic, political, and even social factors. To predict famine requires understanding the complexities of agriculture markets, and Gro Intelligence is trying to use their program to predict famines before they happen. While Gro might be able to churn out an accurate famine prediction two to three months before anyone experiences it, Gro Intelligence needs a plan on how to act on this knowledge.

Even if people are warned in advance of an impending crisis, Sara described the difficulty of getting people to rely on data as opposed to gut feelings. Partnering with groups who have agency to act in the event of a famine prediction and who trust the underlying models, is a challenge for Gro to be fully realized as a tool to prevent and not just detect famine.