We use scientifi models in many different areas of sciences. In meterology and weather prediciton models are made using super computers and very large data sets. Using a model, we can put in the current parameters, and then run the model. The model will then simulate how the weather will change in the given future and therefore we can predict the weather.
Through this comps we were aiming to find a method such that a huge dataset, or a supercomputer was not necessary. The method we used utilizes the nearest neighbors. Look at the image above. The lettered green circles are weather stations have datas from. The black straight arros represent the wind's movement. Then intuitively it would make sense that the future weather of A will look much like B. Like this we use multiple neighbors and predict what the weather wil be like in the future.