How is maize yield related to farm management decisions and household characteristics in Tanzania? Ongoing EPAR research (Report 326) seeks to answer this question, investigating how a farmer who experiences an increase in maize yield from one year to the next is expected to respond in terms of on-farm and off-farm livelihood practices. Will such a farmer specialize increasingly in maize cultivation in order to sell or consume more maize, or hold total production volume constant and shift time or labor resources into other activities? Using an interactive visualization to explore relationships in our data can help us begin to understand the complex dynamics of agricultural production decisions at the farm level.


View Full Screen

The Application

Initially developed to allow EPAR researchers to explore the data and to inform cleaning and trimming decisions, this interactive visualization allows exploration of relationships among variables that represent production and household characteristics, shows how these relationships vary among groups of farmers, and illustrates the sensitivity of the relationships and variable distributions to the treatment of outlier values.

This application is based in the Shiny interactivity package for R, an open-source software environment for statistical computing and graphics. The application development scripts and prepared data are available for download here, and can be used to customize the app to other datasets.

The Data

The World Bank’s Living Standards Measurement Study – Integrated Surveys on Agriculture (LSMS-ISA) is a publicly available household panel survey dataset for seven countries in Sub-Saharan Africa. The survey includes linked plot, household, and community level modules that provide information on crops, livestock, farming practices, and socio-demographics.

Our app displays cleaned data from three panels of the Tanzania National Panel Survey, a LSMS-ISA dataset collected in partnership with the Tanzania National Bureau of Statistics. 

With simple modifications and basic data preparation, the app can visualize any dataset.

By Katie Panhorst Harris and Margaret Beetstra

Summarizing research by Margaret Beetstra, Katie Panhorst Harris, Pierre Biscaye, C. Leigh Anderson, and Travis Reynolds

View a  poster that EPAR exhibited along with this application at the Global Open Data for Agriculture and Nutrition Summit.