I have this set of data that looks like this


I was asked o build a $k$-nearest neighbors algorithm for it which I just finished building. I have this question in regards to the data that I do not understand: Do you notice any spatial or temporal trends in error?

I am not sure how to proceed in answering that question. Any suggestions would be appreciated.


In order to build a 𝑘-nearest neighbors algorithm, you need to define a distance metric that predicts the closest observed data points for a given data point.

For latitude and longitude, great-circle distance.

For time, the distance metric is often days, seconds, or fractions of seconds.

𝑘-nearest neighbors algorithm has no notion of error. Given that there are only a couple of variables, it would be straightforward to do a plot of the when predicting for a new data point.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.