Coming Soon - February 2019
Finalizing new data visualizations
The new, updated GDD site is just the first step in revamping the project’s online presence. Later this year, we’ll be updating our estimates and, accordingly, the way we virtually display them. After all, our goal is to maximize the ways in which people use our data to inform their work, decisions, and even personal habits.
This past year, we’ve had fun learning about the cardinal rules of “data viz” and its critical role in achieving these goals. We’ve talked to experts in fields like economics, sustainability, and public health to get a well-rounded idea of how people of all disciplines will use GDD in the future. We then used these conversations to inform our data viz decisions on things ranging from chart type to filters to color scheme.
Now, we’re in the final phase: building. In the coming months, we’ll be constructing the new visualizations from scratch to make sure they meet the needs of those who will be using them (like you!).
Testing new covariates
To ensure our final, modeled dietary intake estimates are as precise as possible, we have conducted searches to identify data on potential model covariates. Covariates can range from anything like GDP to measures of imports and exports -- really, any additional data about a country which might correlate with its diet trends. This process is especially important for informing what we know about consumption in countries which don't have much (or any) dietary intake data available -- covariates can help to fill knowledge gaps.
Now that we've identified potential covariates, we'll be whittling down our list to the most important ones. To do this, we'll be testing each for inclusion in our model with Spearman's correlation tests at the national level, and then with step-wise linear regressions. Any influential covariates will then be cross-validated for model performance.
For more information about our modeling and covariate methods, check out our Annual Report.