< Home
Final Presentation¶
1.Abstract¶
I used the data, 120 years of Olympic games
This data set contains over 100k athlete datas for 120 years. And it contains the data of country, age, height, weight, year, name.
I wanted to know which athletes would win medals.
I tried to create a machine learning model that estimates whether a medal can be won based on multiple factors.
Initially, we didn't get good results. Since winning medals is a rare case, the learning model began to predict that medals wouldn't be won.
I learned how to refine machine learning models.
Next, I tried to cluster using the K-means method based on weight and height. I plotted everything first, but didn't get good results. Then I narrowed the data down to just the shot put, basketball, and gymnastics events, After that I was able to cluster them clealy.
2.Tried to cluster athlete body size for each sports¶
I wanted to use clustering method for olympic data. Thena I assumed that certain build types are suited for certain sports.
By using k-term clustering method, I tried to cluster athletes.
I wrote about the proccess here.
I could cluster 3 sports using body type. I put height as x-axis, and put weight as y-axis.
Green dots are Gimnastics. Most of gimnastics are short and less weight than other sports.
Yellow dots are showing basket ball player, and blue dots are showing shot put.
I assumed that basket ball player must be high.
Also I assumed that Shot Put athlete must be heavy.
