The secret is out: Machine learning, visualization, and other data science skills are invaluable tools if you want to work in development. But you didn’t need us to tell you that — many of you are already programming away on Python or R, and diving into projects on Kaggle or DataKind.
What many students and young professionals want to learn is how these data science skills are actually being applied in the field. Here are four ways that data science training will continue to be useful to you in your ongoing development career.
The predictive power of machine-learning algorithms
Many organizations in the development sector use a predictive framework to gather evidence to increase their social impact. This method can be used to answer questions such as which geographic areas are most likely to have the most vulnerable households so clients can target where to roll out programs, or which patients are most likely to stop taking their medication to tailor an intervention.
“Even if your role does not directly involve the nuts and bolts of data science, it is useful to know what data visualization can do and how it is realized in the real world.”— Ramie Jacobson
Christy Lazicky, manager at IDinsight, used her data science skills to assess the effects of educational programs on students’ learning outcomes. By predicting results for schools that did not receive the program and comparing those to schools that did, Lazicky was able to identify a rigorous counterfactual to run an impact evaluation.
Another example can be seen in health care, where there is a growing trend to use machine learning — especially as it applies to big data — to predict emergency room visits. Emergency room visits in the United States are expensive, and a financial burden in particular to low-income households.
Researchers are trying to predict preventable emergency room visits so that patients with certain characteristics — disease type, prior and current conditions, etcetera — can solve or mitigate the problems by calling their physicians or nurses instead of going to the ER.
Using visualizations to bridge communication gaps
Even if your role does not directly involve the nuts and bolts of data science, it is useful to know what data visualization can do and how it is realized in the real world.
Lily Li, statistical programmer at the Seattle-based Fred Hutchinson Cancer Research Center, constructs cancer metrics that measure the quality of care in Washington state. She then communicates the results in an application or a website that requires interactive visualization.
More on top emerging skills in development:
In her own words: “Although I am not in charge of coding the data visualization part, I find it useful — and fun — to speak to the data engineers in their language with the ideas of my team — a group of scientists, economists, and data analysts. For example, we used a flow diagram to trace cancer patients’ care pathway after diagnosis. For me, the approach to forming a story and finding the best visualization to tell the story is extremely critical.”
Maria Schwarz, analytics engagement manager at QuantumBlack, manages teams of data scientists and engineers to help businesses, organizations, and governments use their data better at scale. When working with technical teams and decision-makers in business and policy, she has often found that visualization is the most critical step in bridging the communication gap.
For example, Schwarz pointed out that explaining a decision tree without a visual aid is quite difficult, but simply drawing an example tree on a piece of paper or showing an interactive visualization makes the concept very tangible.
Leveraging big data to help governments and international organizations
The words “big data” are everywhere, and development is no exception.
Mar Carpanelli, economist at the Inter-American Development Bank, told us that international organizations such as IDB, the World Bank, and the World Economic Forum are starting to leverage the potential of big data by partnering with Uber, LinkedIn, and others, to complement traditional data sources with real-time, high-frequency, low-cost data for research and policy papers.
Additionally, development startups such as Premise Data and Zenysis are building machine-learning models and visualization dashboards to help governments and international organizations solve some of the hardest challenges facing humanity — from vector surveillance and control of infectious diseases to violent extremism monitoring.
Companies with high development impact potential are also leveraging technology and data science to make decisions. Carpanelli recently partnered with Lori Systems, an African startup disrupting logistics by providing a platform to drive down costs in the cargo transport value chain. She explained that the World Bank estimates that up to 75 percent of the price of goods in Africa comes from logistics, as compared to 6 percent in the U.S. — “so just imagine the huge impact Lori can unlock on quality of life and job creation in Africa by leveraging their platform data to improve efficiency.”
Improving program targeting and product performance
Identifying how to best deploy limited program resources is a common problem for many NGOs and governments in the development sector. In these situations, machine-learning algorithms can help identify which beneficiaries are in most need in order to guide program allocation decisions.
Lazicky cited examples such as predicting which students are most likely to drop out of school, which households are most likely to purchase certain products, and which farms are most likely to have low crop yields.
For health care, Schwarz pointed to an innovation from Nexleaf Analytics that developed a sensor to monitor the temperature in vaccine refrigerators and wirelessly alert health professionals about critical temperature changes. Beyond that, the collected data can be used to analyze the performance of individual fridges — such as to identify chronic failures — as well as to optimize the entire cold chain — such as to understand which fridges work best given the specific conditions of a certain village — to increase vaccine safety and efficiency.
Remember, data science mastery is twice as valuable if you have a solid foundation in econometrics and statistics to help you interpret the data correctly. A graduate program that grounds you in economics while allowing for data science electives can set you on the path to the development career of your dreams.
For more coverage on professional development, visit the Skills for Tomorrow site here.