Introducing the ExpenseVisualizer
It took us almost two years to get there: following a Canadian federal directive, departments had to publish their travel and hospitality expenses. Scattered amongst more than a hundred websites, VisibleGovernment.ca (a Canadian non-profit organization promoting online tools for government transparency) scraped this data and put it into a database. The result: a 22MB SQL database, which just waited for an online tool to unlock its potential.
Publishing data is a necessary step for a more transparent democracy, but this data also has to be accessible enough so that any citizen can use to better understand what is going on in the government. With other governments publishing financial and budget data, we felt it was time to create a tool that would give better, easier access to the information hidden in big spreadsheets and databases.
The ExpenseVisualizer is an online tool that takes advantage of modern web technologies to provide a visual, interactive window on the Canadian Hospitality & Travel Expenses dataset. Our goal was to create a simple tool that would give citizens the opportunity to look at the numbers, discover interesting facts, trends and patterns... and more importantly, to easily share their findings !
The interface displays each department hostpitality & travel expenses as a cumulative area graph (dark blue=hospitality, blue=travel, light blue=hospitality+travel) and offers the following controls:
- options to sort by total expenses, name or deviation (we'll explain this shortly)
- filters to narrow down to a specific set of departments and positions
- a slider to select a specific time frame (ie. all the expenses between January 04 and January 07)
let's find out who are the biggest spender in '05:
The top-3 are Environment Canada (1+M$), Natural Resources Canada (0.75M$) and Veterans Affaiirs Canada (0.72M$). We can also see that 14 departments out of 22 spend less than (0.4M$) -- so why is there such a big difference ? We can already spot that some departments spend more in hopsitality (dark blue) than others, but it's hard to compare, especially for smaller departments.
The ExpenseVisualizer offers a display parameter that can do just that: switch to the "relative" display mode (upper right, just above the graphs) to see the slope for the area in full height, for every department.
Looking at the departments who spend the less, we can see that the slope of the area is much more ragged, especially compared to departments who spend more: is it because they actually spend less, or maybe because they don't report all the expense, or simply because we don't have the data for them ?
By switching to the "monthly" display mode we will see reported expenses per month, and have an opportunity to spot trends and/or missing data.
By switching to the montly display mode, we can already spot interesting things, such as the missing months in the Canadian Grain Commission, National Parole Board, Election Canada and Indian And Northern Affairs departments. Is it because they have "high and low seasons" ? or is it because we have missing data ?
Let's do another test: we filter all the positions, showing only the cumulated expenses for Ambassador, Deputy Minister, Minister, General/Executive Director and Director.
Surprisingly, we see that for almost 50% of the departments, no expenses were reported by any of these top positions -- which raises the question: is this because the departments have no such position ? because they actually had no travel or hospitality expense ? or maybe also simply because there is missing data. This underlines another concern: it is a good step forward to publish the data, but the data also has to be of good quality. As the dataset we are using has been scraped from government website, it is possible that some data is missing, or even erroneous.
As you see, we could already find interesting things just by playing with the interactive visualization parameters. We can also see how valuable it is to have a way to actually view the data: it is much easier to spot quality problems, missing data, possible trends and patterns.
Out intent with ExpenseVisualizer is to show the world how we can leverage web technologies to give citizens more ways to be proactive and gain a better understanding of what's happening around them. We believe that providing visual, interactive tools to explore the data is one of the most efficient way to democratize the data, and let this data be used a new soil for ideas, findings and actions.
So now that you know enough of the ExpenseVisualizer, you can start playing with it and share your findings: simply copy the address (URL) of the visualization and share it with your friends, and of course, embed it in your blog posts to trigger new discussions !
Links:
