Sequence Oriented Process Mining

This repository has the source code for the manuscript Sequence-Oriented Sensitive Analysis for PM2.5 exposure and risk assessment using Interactive Process Mining. Please note that this is a research repository and the code should be improved in future iterations to obtain a whole application. Next, it is described the content of the principal directories.

config

This directory contains all the configuration files needed to run the model. These include the geojson with the I/O ratios and the statistics needed to generate the synthetic population.

data

This directory is subdivided into three:

aq_data

This is the gridded air quality data obtained from an air quality model. In our case, it is generated from the reference air quality stations using bilinear interpolation.

population_data

This is the data generated by the script generator.py, including the citizen's information: id, work, location and epidemiological data.

sequence_data

These are the activities performed by the citizens with the corresponding exposure and risk. It is the Interactive Process Mining input, and the sequences.py script generates it. The v6 labels are the dataset used to compute the KPIs presented in the paper but note that each execution of sequences.py will generate different datasets - with equivalent KPIs

scripts

There are four scripts

generator.py: it takes the configuration files as input and generates the population for the study
sequences.py: takes the gridded air quality data and the population data as input. Then, it generates a CSV with the sequence of activities per citizens

KPI.py: it takes as input the sequence activities and computes the KPIs, plotting them in a gauge diagram

anova.py: it takes as input the sequence activities. It computes the ANOVA analysis with the corresponding plots.

Note that the scenario selection is made through Interactive Process Mining. The PMApp tool for this study is software developed in a previous project, which is not open access, and should be requested from the authors.