Grameen Foundation & Palantir: Partners for Food Security

A Grameen Foundation Community Knowledge worker speaks with a Ugandan farmer

A piece recently published on both the Scientific American and Fast Company’s CoExist websites highlights our most recent work with the Grameen Foundation. We participated in Hacking for Hunger, a first-of-its-kind hackathon held by the Office of Innovation & Development Alliances at USAID, which focused on global food security issues. From the article:

Palantir took 28,000 geo-located soil samples Grameen had taken from across Uganda and combined them with data on soil types, population, income, and other factors. The developers hope the system can also help identify potential disease outbreaks, and help create an alert system for farmers who might be affected.

Read on for more about the data-driven work that Grameen Foundation is doing, pioneering the use of smartphones as two-way information conduits to rural farmers in Uganda, and how Palantir is helping with those efforts.

Grameen Foundation’s CKW Program

Video submission to the Hacking for Hunger hackathon

Grameen Foundation is the charitable organization inspired by the work of Grameen Bank, one of the first organizations to pioneer microfinancing loans for the impoverished in the developing world. The foundation grew out of the success of the bank, and is a separate entity working on lifting people out of poverty. From their website:

Grameen Foundation helps the world’s poorest, especially women, improve their lives and escape poverty by providing them with access to small loans, essential information, and viable business opportunities. Through two of the most effective tools known—small loans and the mobile phone—we work to make a real difference in the lives of those who have been left behind: poor people, especially those living on less than $1.25 per day.

A CKW t-shirt reads, “ASK ME ABOUT FARMING.”

One of their units, the AppLab builds mobile phone applications to leverage the power of cell networks to disseminate information to those who have the least access to it. The Community Knowledge Worker (CKW) Program hires people in Uganda and provides them with Android phones running an open source, custom agricultural information app with an entire data collection platform behind it. These CKWs then travel around Uganda, engaging with rural farmers to provide them with essential information to help them be better farmers. Everything from market pricing, best practices, and disease information is included in the application, and cached for offline access when the CKW is off the grid.

The first time CKWs meet a farmer, they register him or her in their phone, record some brief demographic information (“How many children under 11 do you have?”, “Do your children have shoes? Clothes?”, “What do you use for cooking fuel?”) and start answering questions. Most of the these farmers live outside of the coverage of Ugandan cell networks, but the phones use their GPS satellite signal to record the exact time and location of each query. When the phones return to the grid, all of the data about the queries are uploaded to a central server. The upshot is a perfectly geocoded dataset that maps out a large amount of rural farming done in Uganda.

<img title=”Uganda Land Distribution” src=”/wp-assets/wp-content/uploads/2012/10/uganda-land-distribution.png”</p>
Object Explorer plot of acreage farmed vs. count of farmers (log scale)

Of all the non-profits that the Philanthropy team has engaged with, the Grameen Foundation has some of the finest data we have ever encountered. They know that their data and their two-legged sensor network have value. They are widely regarded to have the best known data on agriculture in Uganda.

Our Work with Grameen Foundation

Object Explorer plot of query distribution among categories

Palantir’s Philanthropy added Grameen Foundation as a partner in mid-2012, having first met them at DataKind’s San Francisco Data Dive.

One of the first workflows that we were able to uncover in their data (highlighted in the video submission) is using query data to track the outbreak of crop and livestock illness. It is possible to see the progression of blights in the query data, which was not foreseen when the CKW data set was being collected. Once the data was imported, it only took about twenty minutes to start mapping out the spread of a chicken blight.

Heatmap of baby chicken blight: Animals Chicken Local Diseases Whitish Diarrhoea In Chicks Less Than 2 Weeks Old And High Death Rate

Public presentation of our work with Grameen Foundation began earlier this year with a presentation by Philanthropy Team Lead Jason Payne to a USAID-sponsored conference on African food security as part of the G8 meeting in Chicago. Here’s a recorded version of that earlier presentation:

Next Stop: Hunger Summit

Our work in the hackathon was a success. We integrated a new dataset collected by the CKWs—the 28,000 soil samples mentioned in the article—and built up some new workflows around the data. We were one of four finalists chosen as part of the hackathon. Grameen Foundation and Palantir will be traveling with USAID to the Iowa Hunger Summit (put on by The World Food Prize Foundation) on October 16th to present our work.