README file 

About the data set 
Title:
Validation workflow and derived comparison data for: Evaluating the accuracy of the EcoservR toolkit for fine-resolution habitat mapping
Sandra Angers-Blondin1, Colm Bowe1,2*, Luke Bentley1, Alex Owusu Amoakoh1,2, Joe Bellis1, Lucy Dowdall1, Noémie Bonnin1, Stamatia Galata1, Hannah Branwood1, Andrew Clark3, Chloe Bellamy4, Hanna Partoft1

1Natural Capital Hub, Department of Biological and Environmental Sciences, Liverpool John Moores University, Liverpool, UK
2Liverpool Research Institute for Climate and Sustainability (LiRICS), Liverpool John Moores University, Liverpool, UK
3Merseyside Environmental Advisory Service (MEAS), Liverpool, UK
4Forest Research, Farnham, England, UK
* Rights holder:
Liverpool John Moores University
* Year of publication:
2026
Description 
This dataset contains the full validation workflow and derived non-spatial comparison data used in the study Evaluating the accuracy of the EcoservR toolkit for fine-resolution habitat mapping. It enables full reproducibility of the accuracy assessment conducted for two study areas in northern England (Merseyside and the North York Moors). The dataset includes R scripts for systematic sampling, spatial intersection of predicted and reference datasets, construction of confusion matrices, calculation of area-weighted accuracy metrics following Olofsson et al. (2014), and generation of summary tables and figures. A non-spatial comparison table is also provided, listing for each validation point the predicted habitat class, reference survey class and adjudicated outcome. Licensed spatial datasets (e.g. Ordnance Survey MasterMap) are not included due to third-party restrictions. However, all analytical steps required to reproduce the validation are fully documented.
Citation:
Angers-Blondin, S., Bowe, C., Bentley, L., Amoakoh, A. O., Bellis, J., Dowdall, L.,  Bonnin, N., Galata, S., Branwood, H., Clark, A., Bellamy, C., & Partoft, H. (2026). Validation workflow and comparison dataset for the EcoservR habitat mapping toolkit. Liverpool John Moores University Research Data Repository. DOI: https://doi.org/10.24377/LJMU.d.00000256

Related publication (in progress):
Angers-Blondin, S., Bowe, C., Bentley, L., Amoakoh, A. O., Bellis, J., Dowdall, L.,  Bonnin, N., Galata, S., Branwood, H., Clark, A., Bellamy, C., & Partoft, H. Evaluating the accuracy of the EcoservR toolkit for fine-resolution habitat mapping. Ecological Solutions and Evidence (under revision).

Contact details 
Principal Investigator:
Colm Bowe
Email: C.Bowe@ljmu.ac.uk
Open Research Team openresearch@ljmu.ac.uk 
Natural Capital Hub, Liverpool John Moores University
Terms of use 
© 2026 Liverpool John Moores University.
This dataset is made available under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence. Users may copy, distribute and adapt the material for any purpose, provided appropriate credit is given.
Project and funding information    
Project title:
Validation of the EcoservR habitat mapping toolkit
Project dates:
Start: January 2023
End: December 2023
Funding:
Internal support from the Natural Capital Hub, Liverpool John Moores University. No external grant number.
Contents  
/explo_scripts/
Exploratory scripts used during method development and testing.
/older_scripts/
Archived earlier versions of analysis scripts retained for version traceability.
.Rhistory
R session history file.
00_clean_data_for_publication.R
Final data-cleaning script used to prepare validation outputs for archiving and publication.
00_clean_MEAS_files.R
Pre-processing script for ecological survey datasets obtained from Merseyside Environmental Advisory Service.
01_accuracy_analysis.R
Primary script implementing the accuracy assessment workflow, including construction of confusion matrices and calculation of accuracy metrics.
01_accuracy_analysis_comparison_old_vs_new_classification.R
Script comparing accuracy results between earlier and revised classification schemes.
02_accuracy_results_caret.R
Auxiliary script for computing classification metrics using the caret package.
03_photo_verification.R
Script supporting manual verification of mismatches using aerial imagery.
04_figures.R
Script for generating manuscript figures and visual summaries of classification performance.
05_accompanying_info_land_stats.R
Script producing land-cover summary statistics for contextual interpretation of results.
06_photo_verif_SAB.R
Script supporting photo-verification workflows for supplementary analyses.
06_photo_verification_visualisation_and_analysis.R
Script for visualising and summarising photo-verification outcomes.
living_england.R
Script used for exploratory integration of Living England datasets during method development.
precision_recall_horizontal_bars.R
Script generating precision–recall bar charts used in manuscript figure preparation.
revised_ecoservR_classification.R
Script implementing the revised EcoservR classification logic used in this study.
revised_ecoservR_module_classif.R
Supporting module script for the revised EcoservR classification workflow.
revised_LCR_basemap.R
Script generating the revised habitat basemap for the Liverpool City Region study area.
revised_NYM_basemap.R
Script generating the revised habitat basemap for the North York Moors study area.
Methods  
The dataset was generated using a deterministic validation workflow implemented in R. Systematic grid-based sampling was used to generate spatially balanced validation points across each study area. Predicted habitat classes from EcoservR outputs were intersected with independent ecological survey polygons derived from Phase 1 field surveys. Where disagreements occurred, high-resolution aerial imagery was used to adjudicate whether mismatches reflected classification error, temporal change, or alternative valid interpretation. Confusion matrices were constructed and area-weighted accuracy metrics were calculated following Olofsson et al. (2014). The archived scripts automate all steps from sampling and spatial intersection through to accuracy estimation and generation of summary tables and figures.
Detailed methodological description is provided inAngers-Blondin, S. et al. (under revision). Evaluating the accuracy of the EcoservR toolkit for fine-resolution habitat mapping. Ecological Solutions and Evidence.

2025