Now-casting Romanian Migration into the United Kingdom by Using Google Search Engine Data
=================================================================================================================

Andreea Avramescu (1), Arkadiusz Wiśniowski (2)(mailto:a.wisniowski@manchester.ac.uk)


(1) Alliance Manchester Business School, University of Manchester, Booth Street West, M15 6PB, Manchester, UK
(2) Department of Social Statistics, School of Social Sciences, University of Manchester, Oxford Road, M13 9PL, Manchester, UK

DOI: https://doi.org/10.5281/zenodo.5508213

The repository contains data and code for reproducing results of the analysis. The file that 
loads data and performs analysis (or loads results) is `output.R`. The simulation is performed 
using `analysis.R`. The simulation uses rstan package. Before running the simulation, please 
read a "readme" note in the `analysis.R` file on package versions and potential issues.

Software used in computations:

- R 3.6.3
- Stan Version 2.21.2, GitRev: 2e1f913d3ca3

Data used:
- International Passenger Survey (age, sex, country of birth), available publicly at: https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/internationalmigration/datasets/internationalpassengersurvey408sexbyagebycountryofbirth 
- Google Trends data, avilable at https://trends.google.co.uk/trends; these data were obtained via R package gtrendsR (https://www.rdocumentation.org/packages/gtrendsR/versions/1.4.8)

> Repository structure


    GT-migr-forecasts/                        
    ├── Code                          <- R the code used for the analysis          
    |   ├── analysis.R                       <- contains setting up of simulations and results saving, setting up data for output                             
    |   ├── data_prep.R                      <- reads in, cleans and prepares data for analysis, also includes plots                            
    |   ├── functions_modelling_plotting.R   <- includes functions for analysis and plotting that are sourced in `analysis.R`  
    |   ├── google_trends_data.R             <- example code to download the Google Trends data using the gtrendsR package
    |   └── outputs.R                        <- contains code for producing all tables and figures presented in the manuscript  
    |
    ├── Data                          <- all data used for the analysis in CSV or XLSX format  
    |    ├── GT_cluster_dictionary.csv              
    |    ├── GT_data                  <- the data used to create the Google Trends Index               
    |    |   ├── Control_Cluster.xlsx       
    |    |   ├── Education_Cluster.xlsx   
    |    |   ├── Employment_Cluster.xlsx       
    |    |   ├── Housing_Cluster.xlsx       
    |    |   ├── Pound_Cluster.xlsx 
    |    |   └── GT Data Compendium
    |    ├── GT_data.csv              <- aggregated data of all keywords used to create the 5 GT clusters            
    |    └── IPS_data.xlsx            <- oficial statistics data between 2004 - 2019
    |
    ├── Graphs                        <- all figures presented in the manuscript in PNG or PDF format                     
    |
    ├── Models                        <- STAN code to build the Autoregressive and Random Walk models
    |    ├── tsmodel011.rds / stan                                              
    |    ├── tsmodel011rw.rds / stan                                            
    |    ├── tsmodel021.rds / stan                                             
    |    └── tsmodel021rw.rds / stan   
    |
    ├── Output                        <- outputs of the analysis (with analysis.R)                                             
    |    ├── results_30.RData                <- models with data starting 2013                       
    |    ├── results_31.RData                <- models with data starting 2012                       
    |    ├── results_32.RData                <- models with data starting 2013 but using differences, not levels                       
    |    └── results_all.RData               <- simulations for all models and years for which forecasts were made            
    │
    ├── LICENSE                       
    └── README.md