ETACE Eco­no­mic Theo­ry and Com­pu­ta­tio­nal Eco­no­mics

Fle­xi­ble Large-​scale Agent Vi­sua­liza­ti­on Li­bra­ry

The Fle­xi­ble Large-​scale Agent Vi­sua­liza­ti­on Li­bra­ry (FLA­Viz) is a data ana­ly­sis and vi­sua­liza­ti­on li­bra­ry de­ve­lo­ped for multi-​agent / agent-​based si­mu­la­ti­ons ge­ne­ra­ted using the FLAME frame­work.

Agent-​based si­mu­la­ti­on mo­dels ty­pi­cal­ly ge­ne­ra­te data that spans mul­ti­ple di­men­si­ons, e.g. pa­ra­me­ter sets, Monte Carlo re­pli­ca­ti­on runs, dif­fe­rent agent types, mul­ti­ple agent in­stances per type, many va­ria­bles per agent, and time pe­ri­ods (ite­ra­ti­ons).

To deal with such a large he­te­ro­gen­ei­ty in the data di­men­si­ons, the data has to be stored as or­ga­ni­zed data sets, to allow for pro­per data ag­gre­ga­ti­on, data fil­te­ring, selec­tion, sli­cing etc. The soft­ware packa­ge FLA­Viz can be used to fil­ter, trans­form, and vi­sua­li­ze time se­ries data that has been stored using mul­ti­ple hier­ar­chi­cal le­vels in the HDF5 file for­mat. Va­rious kinds of plots can be spe­ci­fied, e.g., time se­ries, box plots, scat­ter plots, his­to­grams, and delay plots.

Down­load: FLA­Viz Git­Hub Re­po­si­to­ry