University of Stirling
University of Stirling: DataSTORRE (Stirling Online Repository for Research Data)Not a member yet
118 research outputs found
Sort by
Genetic diversity and ecological success of the invasive riparian plant Mumulus guttatus
No full textImperfect historical records and complex demographic histories present challenges for reconstructing the history of biological invasions. Here we combine historical records, extensive worldwide and genome-wide sampling, and demographic analyses to investigate the global invasion of Mimulus guttatus from North America to Europe and the Southwest Pacific. By sampling 521 plants from 158 native and introduced populations genotyped at >44,000 loci, we determined that invasive M. guttatus was first likely introduced to the British Isles from the Aleutian Islands (Alaska), followed by admixture from multiple parts of the native range. We hypothesise that populations in the British Isles then served as a bridgehead for vanguard invasions worldwide. Our results emphasise the highly admixed nature of introduced M. guttatus and demonstrate the potential of introduced populations to serve as sources of secondary admixture, producing novel hybrids. Unravelling the history of biological invasions provides a starting point to understand how invasive populations adapt to novel environments.VCF file of genotypes of Mimulus guttatu
Elephant event counts, Ruaha-Rungwa 2018-2019
No full textThis dataset was generated by a study investigating how anthropogenic risk affects the active periods of African savanna elephants. The dataset was generated from independent detection events of elephants from camera trap photos obtained during the deployment of four camera grids in the Ruaha-Rungwa ecosystem of Tanzania in the dry seasons of 2018 and 2019: Grid RNP in the core area of Ruaha National Park, Grid MIO in a miombo wilderness zone of Ruaha National Park, Grid MBO in MBOMIPA Wildlife Management Area, Grid RUI in the Rungwa-Ikiri block of Rungwa Game Reserve. Grid RNP was considered a low-risk area for elephants. Grids MIO, MBO and RUI were considered high-risk areas for elephants. We summed the number of independent elephant detection events (column ‘Count’) per camera trap station (column ‘Camera’) by group type (male or female) and diel period (dawn, day, dusk, night). The number of camera trap sampling hours per diel period for each camera trap station are included in the column ‘Hours’. Each camera trap station was either on or off road, and near water (within 1km of water source) or far from water
Dataset for Soil CO2 flux trenching study by Hermans et al.
No full textSoil surface CO2 flux from trenching study to determine autotrophic and heterotrophic CO2 sources in afforested peatland soils. Data are from 4 replicated forestry plots with trenching treatments and litter absence/presence over 2 years. A detailed description of experiment and measurements themselves will be available in the forthcoming paper.Hermans et al Biogeosciences data.csv: CO2 flux (alternatively in units of g CO2 m-2 h-1 or in µmol m-2 s-1). Each measurements specifies the following: Location (4 replicated sites); Date; Day of Year; Days since trenching treatment; Microform where measurement was taken (furrow/old surface/plough throw); litter presence (present/absent); Trenching treatment (trenched/control); Soil temperature; Air temperature; Soil moisture
Datasets for Robust ecological analysis of camera trap data labelled by a machine learning model
No full textThese data accompany Whytock and Swiezewski et al. 2021. Robust ecological analysis of camera trap data labelled by a machine learning model. There are six .csv files in total with information on the species identified in camera trap images and the deployment dates for camera traps. The GPS coordinates of camera traps have been removed to prevent species locations being identified.These data accompany Whytock and Swiezewski et al. 2021. Robust ecological analysis of camera trap data labelled by a machine learning model. There are six .csv files in total with information on the species identified in camera trap images and the deployment dates for camera traps. The GPS coordinates of camera traps have been removed to prevent species locations being identified. Refer to Meta_Whytock_and_Swiezewski_et_al_2021.tx
Generated data and plots for the paper "Exploring the Accuracy -- Energy Trade-off in Machine Learning"
No full textMachine learning accounts for considerable global electricity demand and resulting environmental impact, as training a large deep-learning model produces 284 000kgs of the greenhouse gas carbon dioxide. In recent years, search-based approaches have begun to explore improving software to consume less energy. Machine learning is a particularly strong candidate for this because it is possible to trade off functionality (accuracy) against energy consumption, whereas with many programs functionality is simply a pass-or-fail constraint. We use a grid search and NSGA-II optimisation run to explore hyperparameter configurations for a multilayer perceptron (from scikit-learn) on five classification data sets, considering trade-offs of classification accuracy against training or inference energy (using the PyRAPL) and run times. This includes the generated data of energy, time, accuracy metrics for each dataset, and the full set of corresponding plots. Details for each file are in the enclosed readme.[dataset]_MLP_metrics[_more]_objective1_objective2.pdf
- plots showing the full set of results from the grid search. objective1 is either cross-fold validation accuracy or test data accuracy. objective2 is energy or time on training or testing. ("more" indicates that the larger set of hyperparameter values were used)
[dataset]_MLP_metrics_more_hls_testcpuenergy.pdf
[dataset]_MLP_metrics_more_hls_traingcpuenergy.pdf
- relationship between hidden layer size and cpu energy, for training and testing
[dataset]_MLP_metrics_more_[training|testing]_cputime_vs_cpuenergy.pdf
- relationship between cpu time and energy for training and testing respectively
[dataset]_MLP_metrics_more_training_cpuenergy_vs_testingcpuenergy.pdf
- relationship between training energy and testing energy
pf_aggr_[dataset]_MLP_metrics[_more]_[objective1]_[objective2].csv
- the Pareto fronts from the grid search, aggregated using median values for both objectives
pf_[dataset]_MLP_metrics[_more]_[objective1]_[objective2].csv
- the Pareto fronts from the grid search, no aggregation
pf_[dataset]_MLP_metrics[_more]_[objective1]_[objective2].pdf
- plot of the Pareto front from the grid search
iris-mlp-4-hist-cpu-energy.pdf
- histogram of cpu energy measurements for iris
grid_search_results.zip
- full set of raw data from the grid searches; includes the failed runs where energy is negative
PFs_MLP_Testing.ods
PFs_MLP_Training.ods
- these are the spreadsheets used to show the relationships between hyperparameters and objectives for the Pareto fronts from the grid search
mop.zip
- results from the NSGA-II run on diabetes, with objectives: test accuracy and training energy
- fun_*.txt are the objective values from the Pareto front for each run
- allfronts.csv is the above but concatenated into one file
- var_*.txt are the hyperparam values from the Pareto front for each run
- diabetes*.dat are the 1, 15, and 30th attainment surfaces
attsurface_diabetes_MLP_metrics_more_testacc_trainenergy.pdf
- the figure showing the attainment surface for the mop data abov
Woodland opportunity map of the UK
No full textA raster GIS layer covering the UK, showing percentage of ground potentially available for woodland creation within each 5km grid cell. Areas considered unavailable for woodland creation included higher quality agriculture land and existing woodland (the modification of which could reduce agricultural or timber production, leading to offshoring of emissions), sites designated for nature conservation, priority habitats and peatlands (to avoid perverse outcomes for soil carbon or biodiversity) and existing buildings, infrastructure and archaeological features. Some excluded areas were further buffered to limit possible negative spill-over effects of woodland creation, such as on peatland hydrology, or to reduce predation in sites designated for conservation of wading birds. Full details of the methodology and input layers are available in the layer metadata document.1. Raster GIS layer showing percentage of ground potentially available in each 5km grid cell. 2. Metadata document, outlining the map creation methodology and the input layers used to exclude unsuitable areas.A copy of the GIS file is lodged with RSPB's Open Data platform as a vector shapefile
“Health and welfare of lumpfish in hatchery production and deployed in Scottish salmon cages”
No full textThis is the raw data on the condition and morphometrics of the lumpfish and salmon used and analysed for the manuscript Rey S, Treasurer J, Pattillo C & McAdam BJ (2021) Using model selection to choose a size-based condition index that is consistent with operational welfare indicators. Journal of Fish Biology. https://doi.org/10.1111/jfb.147, as well as the supplementary figures and Rscripts mentioned in the paper.open data for condition,open data for weights and supplementary data (Supplementary Figures 1 and 2 and a supplementary appendix of the R code to recreate all results and figures)
n/a
No full textIndividually based birth and death dates, with early life experiences for elephants in Amboseli Kenya. These are stripped down data; users requiring more information should contact P.C. Lee at [email protected] based birth and death dates, with early life experiences for elephants in Amboseli Kenya. These are stripped down data. This is an Excel spreadsheet. Excel file. Blanks = missing information. -1 = Living (2020)
Counterfactual Reasoning in Children
No full textIncreasing evidence suggests that counterfactual reasoning is involved in false belief reasoning. We add to this evidence by showing that a hallmark error of early counterfactual reasoning appears in a false belief test that makes that error possible. In two experiments we tested 3- to 14-year-olds and found high positive correlations (rs = .56 and rs = .73) between counterfactual and false belief questions. Children were very likely to respond to both questions with the same answer, also committing the same type of error. We discuss different theories and their ability to account for each aspect of our findings and conclude that reasoning about others’ beliefs and actions requires similar cognitive processes as using counterfactual suppositions. Our findings question the explanatory power of standard views, theory theory and simulation theory, in favour of views that explicitly provide for a relationship between false belief reasoning and counterfactual reasoning.Raw data from experimen
Data from: 'Mobile EEG reveals functionally dissociable dynamic processes supporting real-world ambulatory obstacle avoidance: Evidence for early proactive control'
No full textThis dataset is related to the article entitled 'Mobile EEG reveals functionally dissociable dynamic processes supporting real‐world ambulatory obstacle avoidance: Evidence for early proactive control' accepted for publication by the European journal of Neuroscience on January 19th 2021 (https://doi.org/10.1111/ejn.15120). The dataset includes pre processed (mobile) EEG data of 32 healthy participants (age range 19-65) during an obstacle avoidance task in four different conditions (the name of the files match the name of the conditions in the article): free (no obstacles), pre (pre-set adjustment), far (delayed adjustment) and near (early adjustment). Data are segmented around 'OBSTACLE' event.Dataset included the EEG files of 32 healthy participants in 4 different exerimental conditions (32 x 4 for a total of 128 EEG files, .set and .fdt format). Each file is named as 'Sub', number ('_1') and condition ('_free', '_pre', '_far', '_near')