It would be interesting to think about a pallet of colors that harmonize with it.
We will have at least 3 sources of data that we need to show. That is the data related to disease, mostly time series of cases, data related to climate, and the results of the project.
Initially, we will focus on the visualizations of the data and the results. Concerning the data, our main focus will be to visualize the spatio-temporal behavior of dengue in Brazil.
With this aim, we already developed two visualizations:
I) A heatmap with the states of Brazil in y-axis, the quarter of the year in the x-axis, and the color of the heatmap is defined by the incidence (sum of the number of cases notified in the quarter/ population in the state)
The plot above was done in Altair vegas.
II) A map, with an animation (that also works as a slider) that shows the variation of the incidence in the states of Brazil
The plot above was done in plotly. It demands a lot of improvements (I will create an issue specific to it talking about my experience generating it).
I think that we should think about a scheme of color that would work better in this kind of visualization.
Concerning the results, the first one is the forecast of dengue cases. And that it will be the more challenging visualization (at least in my opinion). We will have a lot of models trying to predict the same curve and the question is, how visualize it?
As in the diagram, I think that in the forecast plots, we should address a class of color for each class of models.
We will have something like this, where each line represents a forecast curve: (Below it's a first try of visualization that I did where when hovering the mouse one curve, the output of a forecasted model is highlighted)
By now, that's the logo of the project:
It would be interesting to think about a pallet of colors that harmonize with it.
We will have at least 3 sources of data that we need to show. That is the data related to disease, mostly time series of cases, data related to climate, and the results of the project.
Initially, we will focus on the visualizations of the data and the results. Concerning the data, our main focus will be to visualize the spatio-temporal behavior of dengue in Brazil.
With this aim, we already developed two visualizations:
I) A heatmap with the states of Brazil in y-axis, the quarter of the year in the x-axis, and the color of the heatmap is defined by the incidence (sum of the number of cases notified in the quarter/ population in the state)
The plot above was done in Altair vegas.
II) A map, with an animation (that also works as a slider) that shows the variation of the incidence in the states of Brazil
The plot above was done in plotly. It demands a lot of improvements (I will create an issue specific to it talking about my experience generating it).
I think that we should think about a scheme of color that would work better in this kind of visualization.
Concerning the results, the first one is the forecast of dengue cases. And that it will be the more challenging visualization (at least in my opinion). We will have a lot of models trying to predict the same curve and the question is, how visualize it?
These models can be of specific types, the group starts a classification here: https://drive.google.com/file/d/1mKTL7par-R8d_IaA4HfHexmERuxSo6Dg/view
As in the diagram, I think that in the forecast plots, we should address a class of color for each class of models.
We will have something like this, where each line represents a forecast curve: (Below it's a first try of visualization that I did where when hovering the mouse one curve, the output of a forecasted model is highlighted)