Different graph types may differ in their suitability to support group comparisons, due to the underlying graph schemas. This study examined whether graph schemas are based on perceptual features (i.e., each graph type, e.g., bar or line graph, has its own graph schema) or common invariant structures (i.e., graph types share common schemas). Furthermore, it was of interest which graph type (bar, line, or pie) is optimal for comparing discrete groups. A switching paradigm was used in three experiments. Two graph types were examined at a time (Experiment 1: bar vs. line, Experiment 2: bar vs. pie, Experiment 3: line vs. pie). On each trial, participants received a data graph presenting the data from three groups and were to determine the numerical difference of group A and group B displayed in the graph. We scrutinized whether switching the type of graph from one trial to the next prolonged RTs. The slowing of RTs in switch trials in comparison to trials with only one graph type can indicate to what extent the graph schemas differ. As switch costs were observed in all pairings of graph types, none of the different pairs of graph types tested seems to fully share a common schema. Interestingly, there was tentative evidence for differences in switch costs among different pairings of graph types. Smaller switch costs in Experiment 1 suggested that the graph schemas of bar and line graphs overlap more strongly than those of bar graphs and pie graphs or line graphs and pie graphs. This implies that results were not in line with completely distinct schemas for different graph types either. Taken together, the pattern of results is consistent with a hierarchical view according to which a graph schema consists of parts shared for different graphs and parts that are specific for each graph type. Apart from investigating graph schemas, the study provided evidence for performance differences among graph types. We found that bar graphs yielded the fastest group comparisons compared to line graphs and pie graphs, suggesting that they are the most suitable when used to compare discrete groups.

We have generated sets of the problem instances obtained by using different pseudo-random methods to generate the graphs. The order and the size of an instances were generated randomly using function random() within the respective ranges. Each new edge was added in between two yet non-adjacent vertices randomly until the corresponding size was attained. This dataset is an extension of the Random Graph dataset available at https://data.mendeley.com/datasets/rr5bkj6dw5/8.

This repository is part of the Ph.D. thesis of Isabelle M. van Schilt, Delft University of Technology.

This repository presents a complex stylized supply chain discrete event simulation model of a counterfeit Personal Protective Equipment (PPE) supply chain. Additionally, this repository presents scripts for automatically generating a discrete event simulation model from a networkx graph. The generation of a large set of randomly generated networkx graphs based on real-world data, which can be automatically ran as a simulation model, is also presented. This contributes to research on structural uncertainty in models. This code is an extension of the Master Thesis of Bruno Hermans , Delft University of Technology.

The simulation models are developed in pydsol-core and pydsol-model . For the real-world data, we use the repository port_data_graphs to create various graph structures based on open-source data.

This data set contains 194778 quasireaction subgraphs extracted from CHO transition networks with 2-6 non-hydrogen atoms (CxHyOz, 2 <= x + z <= 6).

The complete table of subgraphs (including file locations) is in CHO-6-atoms-subgraphs.csv file. The subgraphs are in GraphML format (http://graphml.graphdrawing.org) and are compressed using bzip2. All subgraphs are undirected and unweighted. The reactant and product nodes (initial and final) are labeled in the "type" node attribute. The nodes are represented as multi-molecule SMILES strings. The edges are labeled by the reaction rules in SMARTS representation. The forward and backward reading of the SMARTS string should be considered equivalent.

The generation and analysis of this data set is described in
D. Rappoport, Statistics and Bias-Free Sampling of Reaction Mechanisms from Reaction Network Models, 2023, submitted. Preprint at ChemrXiv, DOI: 10.26434/chemrxiv-2023-wltcr

Simulation parameters
- CHO networks constructed using polar bond break/bond formation rule set for CHO.
- High-energy nodes were excluded using the following rules:
(i) more than 3 rings, (ii) triple and allene bonds in rings, (iii) double bonds at
bridge atoms,(iv) double bonds in fused 3-membered rings.
- Neutral nodes were defined as containing only neutral molecules.
- Shortest path lengths were determined for all pairs of neutral nodes.
- Pairs of neutral nodes with shortest-path length > 8 were excluded.
- Additionally, pairs of neutral nodes connected only by shortest paths passing through
additional neutral nodes (reducible paths) were excluded.

For background and additional details, see paper above.

This article proposes a graphical model that handles mixed-type, multi-group data. The motivation for such a model originates from real-world observational data, which often contain groups of samples obtained under heterogeneous conditions in space and time, potentially resulting in differences in network structure among groups. Therefore, the iid assumption is unrealistic, and fitting a single graphical model on all data results in a network that does not accurately represent the between group differences. In addition, real-world observational data is typically of mixed discrete-and-continuous type, violating the Gaussian assumption that is typical of graphical models, which leads to the model being unable to adequately recover the underlying graph structure. Both these problems are solved by fitting a different graph for each group, applying the fused group penalty to fuse similar graphs together and by treating the observed data as transformed latent Gaussian data, respectively. The proposed model outperforms related models on learning partial correlations in a simulation study. Finally, the proposed model is applied on real on-farm maize yield data, showcasing the added value of the proposed method in generating new production-ecological hypotheses. An R package containing the proposed methodology can be found on https://CRAN.R-project.org/package=heteromixgm. Supplementary materials for this article are available online.

Data for graph (E) is given in a sheet containing an image identifier and fluorescence intensity measurements of Laminin antibody staining (y axis) in each genotype (x axis). (XLSX)