We used Census data of U.S government and visualized it using d3. We implemented bar graphs, choropleth for geolocation based data, pie charts, and scatter plots. Our goal was to visualize the data to the county level in which we compared two different variables like poverty level and educational attainment of that particular county. We had data of following variables which we plotted on the using d3.
Our next step was to find appropriate charts using which data would make sense. We brainstormed different ideas for deciding on the types of representations (e.g., charts) to use in our visualization. Our choices were based on data types which we are trying to depict. For instance, to show age distributions, we decided using histogram (as a bar chart), whereas for percentages/ratios a pie (or a doughnut) chart was our choice.
We had to show population of all the counties in US which had greater than 65K population. Best way was to visulize this data was using chloropleth. Here you can see the density of the color changes according to the population density. Also we provided a color lengend which provided the range of the population. Here grey area are the counties which had less than 65K population
Use of scatter plot was an obvious choice as we wanted to find the relation between two variables of all the counties county. So in total, we found where does a particular county stand with respect to the two different variables. The user has the option to select any variable from the 12 different variables.
For ease of interaction, we provided a feature where the user would click on one county from the choropleth and based on the county selected other variables of the county would be displayed using following charts.
After brainstorming which chart would be used for which data, we divided different chart and developed them differently. When we integrated the chart, the whole picture looked disconnected. So we came up with the idea of the dashboard, where the user would have some interactions to perform in order to make the sense out of the presented charts. Our main challenge was to arrange the charts in a manner which would be aesthetically pleasing and deliver a good user experience. To keep things simple, we decided that the user should be given a single screen for filtering the data through selecting a specific county to find some new patterns and insights.
You can access the live project here.