Predicting Geolocation of Instagram Images - Nathaniel Flick, Brian Vincent, Nikhil Loney, Ajay Kulkarni
-
We have built a model to predict the location of an Instagram image based on its textual tags
-
We used K-Means clustering to divide the world into a set of discrete cells, each of which formed a class
-
Using these classes, we build a Naive Bayes classifier to predict the most likely class for a given input sample
-
We also developed a web page to visualize the results
-
Prediction accuracy ~65% (based on 5-fold cross validation)