In this project I have aggregated four different ways to illustrate data provided by Google’s public domain. Google has taken data from over twenty countries and recorded the frequency of certain search terms. Google then compared this data to that of traditional flu surveillance systems, i.e. the Center for Disease Control (CDC) in the United States. From this pattern, Google creates an actual flu incidence estimate based on search term data. Google believes that their data is significant for health officials because Google is able to predict daily trends whereas surveillance systems do not have access to such up to date data in order to respond quickly to early outbreaks.
In my first visualization, Google compares their estimate based on search terms to the actual incidence reported by the CDC in the United States. This line graph follows five years worth of data. The data is presented in contrasting colors with a legend on the side of the graph. The X axis demarcates time in years, and the Y axis is measured in cases. Google did not delineate whether the cases are by the thousand or otherwise. Using the assumption that in a population of 300 million, the maximum listed number of 6,885 probably is measured in the thousands. For a timeline, I think the line graph effectively demonstrates information. Based on the timeline, we can see that Google’s estimate based on search terms closely follows the actual incidence of the flu. The peaks make it clear to visualize flu seasons. Since Google accurately demonstrates that flu search terms mimic actual outbreaks, a viewer can trust future estimates based on Google data. For single country cases, the line graph in which actual incidence and estimated incidence is sufficient to convey a message.
I chose to use Google’s animated world map of flu estimates based on search data for my second visualization. In the animation, the map is marked with national borders. Each country for which data is available has a bubble. An asterisk marks countries that did not have data available for the particular year in the timeline, and a bubble appears when data is used. Data is demonstrated by playing the timeline over five years worth of data. The bubbles corresponding to each country subsequently enlarge or shrink based on the magnitude of the estimate. Users can highlight a country, and its bubble will appear in a different color during the video. In the link that I have provided I highlighted the United States. Users can also zoom in on regions, and even further break down data by states within certain countries. If only one country is highlighted, it is easy to visualize comparisons and track the spread of the flu. Highlighted all the nations makes data less clear. I think the option to put the data in motion is very innovative, and makes an important statement about outbreak movement and flu seasons.
In my third visualization, I have more data in motion . This time bars represent the countries. Here I chose to select every country so that each had a distinct color and a label. The play button moves data through time. The data is normalized to make the bars comparable between countries. The baseline is set as the usual amount of times flu terms are searched, over several years. Therefore activity in the graph demonstrates search frequencies different from the baseline level. That way the user can see a comparable difference between two countries’ flue incidence. Playing the graph makes bars taller, shorter, or on the other side of their baseline. Here the Y axis ranks countries in descending order, and the X axis moves through months. As data becomes available corresponding countries are added in the motion. It is interesting to see more distinctly how countries compare to each other. However, highlighting all the countries obscures the labels, and the multitude of colors can be distracting. In the motion map, bubbles do not provide a concrete measurement, but they better illustrate the spread of the flu.
For the final visualization I created a bubble chart from Many Eyes. I downloaded the raw data from Google. After I organized it, I uploaded the data set for one week’s worth of flu estimate, that of the week February 5th, 2012. Each nation has a corresponding number. The bubble chart delineates each country with its own bubble. The circumference of each bubble is based on the value Google provided as the estimate of incidence based on flu search terms. The color for each bubble is arbitrary. Users are able to interact on a limited basis in order to find the label for the smaller magnitude bubbles. The larger bubbles have country names, and the largest have both country names and values. I thought it was interesting to see that some countries in the southern hemisphere, like Chile, have very low values as it is currently summer there. The bubble chart is not as dynamic as the motion map or bar graph, but it accurately compares each country to the other. I think it is pleasing to the eye and cleanly organized. However since there is no time parameter, as it is only capable of demonstrating data in a static time frame, the bubble chart is not as effective in representing flu spread.
In conclusion, the data Google presents is very useful for institutions for the CDC and other countries’ respective systems. Each graph demonstrates the data differently. The line graph proves that Google’s estimates closely mimic the actual incidence rate. The animated world map depicts the spread of flu outbreaks geographically and over time. The bar maps provides a useful comparison between countries over time. The bubble chart takes a snapshot in time and also compares between countries. No one graph is most effective, but some illustrate certain points better than others.