Simulation results
output.RmdResults and analysis
In future iterations of the package, I intend to add functions to help summarize and display this output in useful ways. If you have thoughts about what kinds of things may be helpful here, please get in touch!
Currently, BayesACT produces three types of output:
CSV output
Likely of most use for analyzing results is the csv output. This is
saved to the directory specified in run_bayesact(). Read
this in with read.csv() or readr::read_csv().
This csv file contains one row per actor turn, and reports a number of
different statistics about the state of the interaction on each
turn.
Here are the results created by the example shown on the simulation setup and run help page It’s easy to see here that many of the identities and behaviors chosen do not make much sense in context. Subsetting the dictionaries to identities that only belong to a relevant institution may help here.
results <- read.csv2("/path/to/output/readme_simfile.csv", sep = ",", header = TRUE)
head(results[,1:6])| iteration | dyad.0 | dyad.1 | AGENT..agent.name | AGENT..agent.ids | AGENT..agent.id.probabilities |
|---|---|---|---|---|---|
| 0 | 0 | 1 | Sally | teacher | 1 |
| 1 | 1 | 0 | Reem | genius,non_smoker,conservative,consultant,doctor,adult | 0.551,0.224,0.084,0.072,0.068,0.001 |
| 2 | 0 | 1 | Sally | teacher,skilled_worker,doctor,fiance_male,genius | 0.942,0.048,0.007,0.002,0.001 |
| 3 | 1 | 0 | Reem | doctor,assistant,non_smoker,genius,adult | 0.708,0.27,0.019,0.002,0.001 |
| 4 | 0 | 1 | Sally | teacher,skilled_worker | 0.985,0.015 |
| 5 | 1 | 0 | Reem | genius,consultant,non_smoker,doctor,organizer,conservative,surgeon,taxpayer | 0.334,0.295,0.174,0.131,0.039,0.014,0.012,0.001 |