BibTeX
@MISC{Lim_quantifyingthe,
author = {Hyeeun Lim and Nupur Bhatnagar},
title = {Quantifying the Predictability of a Personal Place},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Users visit different places and some places are important to them as compared to others. The rating level was given by the user to rate the importance level of a place. This project involves studying the variable patterns of user's commuting behavior and builds a classification model that could serve as a descriptive as well as predictive model. We have defined four categories of places based on the explanatory variables like frequency of visits by the users and the time spend or duration by them. Clustering of different places in order to assign them a class label was done by using clustering algorithms like K mean and DBSCAN. The classification model was used as predictive model to quantify prediction of a place given a combination of explanatory variables like frequency and duration. Different classifiers like KNN, Naïve Bayes, and Decision Trees were used to build the predictive model. We have evaluated our model using different accuracy and precision parameters Like F measure and based on this compared the performance of different classifiers. We concluded that KNN worked best for our predictive model and we could distinguish between different places labeled by our classification model. Keyword: Location-aware, GPS, classification, descriptive modeling, clustering, predictive modeling, frequently visited places, important places
Keyphrases
predictive model different place classification model personal place different classifier explanatory variable different accuracy decision tree important place rating level predictive modeling variable pattern importance level visited place class label precision parameter descriptive modeling time spend
