READ ME File for 'Dataset for Modelling railway station choice: can probabilistic catchments improve demand forecasts for new stations?'

Dataset DOI: 10.5258/SOTON/D0825


ReadMe Author: Marcus Young, University of Southampton

This dataset supports the following thesis:

Young, M. A. (2019). Modelling railway station choice: can probabilistic catchments improve demand forecasts for new stations?. (Unpublished doctoral dissertation). University of Southampton (Transportation Research Group).

DESCRIPTION OF THE DATA

This dataset includes: model command files and calibration results for the station choice and trip end models reported in Chapters 6 and 7 of the thesis; all tables included in the thesis provided in MS Excel format; command files used for the cross-validation exercise for station choice and trip end models and summary results.

The passenger survey data used to calibrate the station choice models reported in this thesis were obtained from the Welsh Government and the Transport Scotland LATIS Service. These datasets are subject to licence restrictions and cannot be made openly available. Transport Scotland, who carried out the original collection and analysis of these data, bears no responsibility for their further analysis or interpretation.

The zip file, 'supporting data.zip', contains the following:

/NLOGIT models
	/latis
		/latis_thres_mnl.lim
		/latis_thres_rpl.lim
	/wg
		/mnl.lim
		/rpl.lim
	/combined
		/combined_mnl_not_staffing.lim
		/combined_mnl_main_added_not_staffing.lim

These NLOGIT files contain the model commands for creating additional variables and running the MNL and RPL models for the LATIS and WG datasets and the MNL models for the COMBINED dataset (See Chapter 6 of the thesis). These are plain text files.
	
/NLOGIT models
	/latis
		/latis_thres_mnl.lim
		/latis_thres_rpl.lim
	/wg	
		/wg_thres_mnl.lim
		/wg_thresl_rpl.lim
	/combined
		/output_combined_mnl_not_staffing.lim
		/output_combined_main_added_mnl_not_staffing.lim
	
Files containing the model calibration results as produced by NLOGIT during the model runs for MNL and RPL models for the LATIS and WG datasets and the MNL models for the Combined dataset. These contain the results of all model runs. For reasons of brevity only a selection of these models are reported in Chapter 6 of the thesis. These are plain text files.

/NLOGIT models
	/latis
		/sumstats_all.csv
		/sumstats_choice.csv
	wg/
		/sumstats_all.csv
		/sumstats_choice.csv
		
Summary statistics for the predictor variables used in the LATIS and WG station choice models.  These are summarized for all alternatives in the model and for only the chosen alternatives in the model. These are reported in Chapter 6 of the thesis (Tables 6.3 and 6.4). These are plain text CSV files.

/NLOGIT models
	/combined
		/cross-validation procedure
		/combined_mnl_not_staffing.lim
		
The code used for the NLOGIT procedure used in the cross validation exercise for the combined model.

/NLOGIT models
	/combined
		/cross-validation procedure
			/te19
				/cvrep_summary.csv
			/te24
				/cvrep_summary.csv
				
These files contain a summary of the predictive performance difference measure (%) for the 10-fold cross validation that was repeated 10 times for the combined dataset station choice models TE19 and TE24. See Section 6.7.4.1 of the thesis. Plain text CSV files.


-----------------------------------

/Trip end models - R
	/run_models_ews
	
Commands used to run the trip end models in R using lm(). Also code for generating the standardized residuals and carrying out the cross validation exercise and plotting results. R file (plain text). See Chapter 7.

/Trip end models - R
	/model-output.txt
	
Model output from the trip end models run using lm() in R. Plain text files. See Chapter 7.

/Trip end models - R
	/cvsummary.csv
	
Summary of mean square for 10-fold cross validation that was repeated 10 times for model 9, and average mean square for each repeat. Plain text CSV file. See Section 7.6.3 of the thesis.


-----------------------------------

/Tables
	/tables.xlsx
	
The original tables that appear in the thesis. These are in MS Excel format and were converted to LATEX for inclusion in the thesis using the Excel2LaTeX add in. See: https://ctan.org/pkg/excel2latex?lang=en.

-----------------------------------


GEOGRAPHIC LOCATION OF DATA COLLECTION

University of Southampton, U.K.

DATE OF DATA COLLECTION

Data analysis took place 10/2014 to 02/2019.

RELATED PROJECTS/FUNDERS:

ESPRC DTG Grant EP/M50662X/1


DATE THAT THE FILE WAS CREATED: 03/2019

-----------------------------------