Visualization Woes

Happy Monday!

Last week, the Westside Communities Alliance Data Dashboard team finally finished cleaning up data! We have now officially moved onto the Data Visualization Phase.

We’ve identified three sections that we would want on the public safety dashboard:

1. a heat map of the crime data to see where crime is located and lay it with
   1. community assets (e.g. churches, libraries, schools) to see what spots the new assets could move into and to coordinate existing assets,
   2. locations of code violations, particularly, vacant houses, to see if crimes are occuring at those locations;
2. correlation plots to show
   1. the correlation between a crime statistic (e.g. violent crime rates, auto theft rates) and a census statistic (e.g. percent of seniors, school enrollment)
   2. the differences in the strengths of correlations between different pairs of factors (e.g. youths and crimes vs. vacant houses and violent crimes)
3. summary to display the community profile.

We’ve created a few different visualizations for each section to provide options to the shareholders to see which ones get the point across in the most straightforward way.

In general, the yellow – red points represent crime density, while the green points represent vacant housing. With the opening of a new stadium in the Westside, money is being allocated to remove vacant housing in these neighborhoods. It is important that the money is being targeted at vacant housing where crimes occur. From our map, it is clear that there are some areas where crime and vacant housing is correlated spatially, however this is not always the case. While there are issues with the data involving selection effects in areas with a high density of vacant houses, it is safe to say that vacant houses in areas with high criminal activity should be preferentially targeted for demolition over vacant houses in areas with lower crime density.

 

For the correlation plots, we made a scatterplot and a slope graph (a slope graph would show a high correlation if all the slopes in the graph were going in the same direction) and are fine-tuning it.

To compare different correlations, we are working on a number of visualizations. In the one above, the different shades of the bars indicate different crime related statistic. You can filter out what crime statistic you would like to see and hover over the bars to see what corresponds with what. So here, occupied housing has a small negative correlation with % aggravated assault in Atlanta overall, small positive correlation with larceny in Atlanta overall but a strong positive correlation with larceny strictly on the Westside.

In addition to coming up with more ways to illustrate the data for the first two sections, the summary part is where we are focusing on currently to make a concrete prototype.

We’re meeting with the public safety committee of Vine City this Wednesday and are excited to hear their critique to improve our tool!