Search Target

Results of the Winter Olympic Medal Predictions: Using Data and Changing Factors

The closing ceremonies of the Sochi Winter Olympics took place on Sunday, February 23rd but before we say goodbye to the Winter Olympics, let's take a look back at the medal predictions of Dan and Tim Graettinger and see how accurate their data model predictions were.

The Graettinger's model had some accuracy but it was not completely right. Below are two bar graphs, the first showing the Graettinger's predictions while the second shows the actual medal counts of each country.

Graettinger's Olympic Prediction Model

Graettinger's Olympic Prediction Model

Medals Won at the Sochi Olympics

Medals Won at the Sochi Olympics

When looking at the Graettinger's data model, it is important to keep in mind that this is a newly developed model. The brothers only had this idea of predicting medals after the last Winter Olympics in 2010. They built their model based on data they collected, practiced their model during the 2012 Summer Olympics and tested it on previous Olympics that have taken place. Because their model is new, it was more likely that inconsistences with their predictions would occur. In their original article, the brothers acknowledged that their data model would not predict with 100% accuracy but they wanted to see how close they could get to 100%. They are aware of some outlying factors that alter the accuracy of their predictions such as the popularity of a particular sport in a country.

Even with their predictions being a bit off, they were accurate about some of their outlying factors. The Graettinger's predicted that the country hosting the Olympics would over perform their prediction which was true for Russia who walked away from the games with 33 medals, the most at the Sochi Olympics. In addition, the brothers recognized that Austria, Norway and Canada always seem to perform well at the Winter Olympics and would over perform what the model predicts. This was true as Austria was predicted to win 10 medals and instead won 17. Norway was predicted to win 16 and actually won 26 while Canada was predicted to win 18 and won 25.

When thinking about data and data models, it is important to continuously analyze your data and take different factors into account. Maybe some of the information the Graettingers used was not as accurate as other data or maybe there was something else to take into account in order to make their prediction model more accurate. When looking at data, we must be critical of our outcomes and work to improve them in order to make more accurate predictions each time.

 Becoming an Olympian: Making Predictions Based on Data




As the Winter Olympics continue, we tune in to each event and become excited to see how each athlete will perform and which country will leave Sochi with the most medals. Some people, however, already have a good idea of which countries will be victorious. During the 2012 Summer Olympics, an economics professor at Colorado State, Dan Johnson, predicted medal winners by using economics. Nearly two years later, two brothers, Dan and Tim Graettinger, have emerged with different data to predict the 2014 Winter Olympic medal winners.

After the last Winter Olympics in Vancouver, Dan and Tim Graettinger began wondering why certain countries win medals while others do not. As data miners for Discovery Corps, Inc., they constantly analyze data from different perspectives and summarize it to gain useful information. Data mining allowed them to use past information to predict the future on a daily basis and so they considered how data mining could allow them to predict a country's amount of medal wins at the Olympics and how close their predictions could be to the actual number of medals awarded.

When trying to predict winners, we might think that the most important data to analyze would be individual athletes, their achievements in the past, and their performance in a sport. The Graettinger brothers disregarded the athletes altogether when making their predictions and collected as many different pieces of data about a country as possible. After running many regression analyses, the Graettinger's discovered that the four variables to best predict the number of medal wins per country were geographic size, GDP per capita, value of exports, and latitude of the Nation's capital.

The most essential piece of the puzzle for their predictions, and the most surprising, was using a country's medal count from the previous Summer Olympics. The Graettinger's figured out that no nation won a medal in the Winter Olympics without winning at least one medal in the previous Summer Olympics. Keep in mind though that the Graettinger's found that the country hosting the Olympics often over performed their predictions.

So who will come out on top? The Graettinger's have predicted that the United States will come home with the most medals, 29, followed by Germany with 23 and China with 22 medals. We will check in at the conclusion of the Olympics to see if their predictions were right.

To leave you with something to snack on, how can your department use data to predict the future? What information can make the biggest impact? Record and follow your data to see common trends that could help you in the long run and allow you to become an Olympian of Student Affairs.

Last Updated: 11/11/16