Technical notes on a behavioral analysis of Roman sites in rural Tuscany.
Information may also be found at https://volweb.utk.edu/~scolli46/rppsf.html.
The code below depends on the following csv files which contain the semantic graph, as the original graph was composed in Gephi and exported:
- dataset.csv - table of small finds listed by site, phase, and definition, by count and weight.
- nodes.csv - the nodes of the network.
- edges.csv - the links of the network.
- sfcodes.csv - a list of codes used for the RPP small finds assemblage, derived from rppsfnodes.csv.
- rppsfcodes.csv - definitions and metrics of each artifact-type.
- qmethod.csv - quantitative metric used for each artifact-type (c: count, w: weight, p: presence/absence).
A more compact version of the network is available in JSON:
- rppsf.json - full semantic graph with nodes and links.
The following scripts compute an index of a function or behavior, as indicated on the semantic graph above, for further multivariate analysis:
- rppsf-ontology.py - associates artifact-types with behaviors of interest and generates a set estimates of the intensity of that behavior as an index from 0 to 1 through random subsampling.
- https://github.com/scollinselliott/rppsf/blob/master/R/indices.r - used to produce plots; resamples distributions generated from the previous script to evaluate credible intervals.
- https://github.com/scollinselliott/rppsf/blob/master/R/mds.r - performs non-metric multidimensional scaling on the modes (argmax) of behavioral indices. Creates a distance matrix (Euclidean) and uses Kruskal's nMDA (iso.mda).
- https://github.com/scollinselliott/rppsf/blob/master/R/rjc-la-dist.r - calculates a measure of overall intersite variation or differentiation throught the average distance among sites' behavioral indices, for the Late Republican / Julio-Claudian and Late Antique periods.
These files can also be downloaded in a single folder from https://volweb.utk.edu/~scolli46/rppsf/rppsf.zip.