I had the pleasure of presenting with Keera Pullman at Ozri 2012 the new analysis capabilities of ArcGIS 10.1. There are 85 new tools and 117 new parameters that have been incorporated into the ArcToolbox to make your analysis workflows more valuable and more efficient.
We worked through an analytical process investigating the potential variables and factors that affect petrol pricing in the Sydney area. To do this, we implemented three new spatial statistics tools – Grouping Analysis, Areal Interpolation and Exploratory Regression.
Our analysis involved taking petrol station locations with the corresponding prices from a spreadsheet and geocoding it using the ArcGIS Online World Address Locator, used directly within the ArcGIS Desktop interface. After visualising the data, we ran Grouping Analysis on the data set in order to differentiate between pricing groups around the city. With this tool, we were able to maximise the similarities between the prices of petrol stations over geographic space, as well as distinguish as much as possible between the price ranges between the groups. We noticed from this analysis that the prices of petrol were the highest near the Sydney CBD and the North Shore, and lowest in the more inland neighbourhoods.
From here, we can start to query why this pattern exists, and whether or not the proximity to certain points of interest or features causes a significant change in the pricing of petrol. We used drive time buffers (generated directly from a Python script no less) to accumulate our data. Each petrol station was associated with the surrounding features within a 3-minute drive such as schools, other competitors, etc. We also used areal interpolation (a process which reaggregates data from census blocks to our drive time buffers through an interpolative process) to look at potential factors such as median household incomes.
Our final step involved Exploratory Regression. This process takes all the different variables that we’ve associated previously with each petrol station to understand the driving influences on the setting of a suitable petrol price for each neighbourhood. We noticed from our results that things such as the proximity to the airport, or other transportation hubs were significantly correlated with the prices.
Our analytical workflow shows that it does save you money to take into account the neighbourhood characteristics you are in, and to shop around for a better price. Keera was also able to share all of her analyses and outputs in a geoprocessing package to ArcGIS Online, so that anyone else who is interested in running a similar analysis can use her methodology without having to recreate everything from scratch!
I ended our presentation with a list of five of my favourite new improvements to ArcGIS Desktop 10.1. These included:
- Being able to search for data or projections not only through keywords but by extent as well;
- Enabling Editor Tracking in your file geodatabase to automate documenting the creation and modifications to features by users directly in the attribute table;
- Creating a point layer file with photos as geodatabase attachments directly from a folder with geotagged photos;
- Working with KML layers in ArcGIS Desktop;
- Dynamic legends that intelligently display only the legend items available in the current view – excellent for those data driven pages.
I am extremely excited about how this new functionality can make your lives easier and improve the stories you want to tell with your data.