In 2018 the Center for Computational Science sponsored the VizUM Visualization Competition. The focus was on civic engagement and the tag line was; "Your City, Your Neighborhood and You". I decided to enter and invited (AKA coerced) two of my work mates at UM to join me. Suri and Carmel, so we entered as "Team UM Analytics". The goal was to gain more skills with visualization and specifically Python. Our first attempt with GEOPandas was a failure so we switched to Power BI and tried a Sand Dance viz. At the first checkpoint our Sand Dance viz was completely outclassed by the R-studio and Tableau presentations of the other teams. ( Did I mention that BI was completely OUTCLASSED). After that embarrasement we went back to Python and finally got Seaborn to work. GeoPandas was replaced by ArcGIS and Folium and we finally started to make some progress.
This is a link to the GIT live site.
We decided to host our results on Google pages because at the time we did not have GIT hub up and running. Here is the Google pages site:
Google UM_VIZ SITE
We found that the average speed for the accidents in this dataset was higher on Sundays. This is a countplot with error bars in Seaborn.
The average speed was higher from 3AM to 6AM
Males tended towards higher speeds. Seaborn violin plot.
We used ArcGIS near function to calculate the number of accidents within 20 meters of an intersection, then exported these results as an excel sheet and plotted them with seaborn.
The speed tended to vary depending on the neighborhood. Here we used ArcGIS to create a neghborhood category by using the spatial join tool, then we created a dot plot of speeds by neigborhood.
We compared drivers and pedestrian ages by gender and found more female pedestrians(turquoise plot) aged 60-80 were involved in accidents than female drivers but more male pedestrians (tan plot) 18-20 were involved in accidents. Also male drivers in their 40's (purple plot) had more accidents and female drivers (blue plot) in their 20's had more accidents.
