University of Twente Student Theses
The Forecasting-Ability of Google Search Frequencies for Predicting Crime in the Netherlands
Mueller, S.-A. (2019) The Forecasting-Ability of Google Search Frequencies for Predicting Crime in the Netherlands.
PDF
829kB |
Abstract: | Being able to predict the development of crime includes several advantages. In the past, some interesting approaches have been already made to evaluate other factors that are useful for predicting the development of crime besides crime rates themselves. The aim of this paper is it to explore one of these factors by answering the question whether Google search frequencies (GSFs) of terms that display the intention to commit a crime would be able to enhance the accuracy of a quantitative model in forecasting future crime development. To explore these questions, data collected from the platform Google Trends or the data portal of the Central Bureau for Statistics (CBS) are used to create three different models: 1) a linear regression model, 2) an AR(3) model, and 3) an ARIMA(3,1,3) model. The results show that GSFs, while not necessarily lessening the forecasting ability of a model, do not increase the accuracy of the model compared to a model without GSFs. Although the results in this study are non-significant, further investigations are highly recommended. Keywords: Crime, Time-series analysis, Google search frequencies (GSFs), Forecasting, Autoregressive, ARIMA |
Item Type: | Essay (Master) |
Faculty: | BMS: Behavioural, Management and Social Sciences |
Subject: | 77 psychology |
Programme: | Psychology MSc (66604) |
Link to this item: | https://purl.utwente.nl/essays/78292 |
Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Repository Staff Only: item control page