4.3 Output Spatial Patterns and Metrics
4.3.4 Exporting Outputs
Spatial patterns, accessibility and segregation metrics produced by the model can be exported for communication purposes and further analysis on specialised statistical and GIS software packages. Simple and open file formats were chosen for that purpose, described as follows.
Flow Metrics
The spatial patterns can be exported in the form of videos or of a series of snapshots showing the simulation dynamically unfolding. Those visual outputs can be visualised and exported directly from the model’s interface. The flow metrics containing the aggregated flow of agents on each cell, both total and by population group, can be exported as rasters in ASCII grid format. Those files are simple text files which can be opened in most GIS software packages, where further analyses can be carried out. The artificial trajectories simulated by the model can also be exported to the routes.csv file. The structure of that file can be seen in table 4.6.
Table 4.6: Routes table structure (routes.csv). Field Description Data type agent id Agent’s unique identifier Integer route id Route’s unique identifier Integer x Cell’s x coordinate Integer y Cell’s y coordinate Integer order Cell’s order in route Integer
Accessibility
Accessibility and travel statistics of each individual agent can be exported to the accessibility.csv file, detailed in table 4.7.
Table 4.7: Individual accessibility table structure (accessibility.csv).
Field Description Data type
agent id Agent’s unique identifier Integer agent group Agent’s group identifier Integer transport mode Transport mode String
movement speed Agent’s speed Floating point origin zone Origin zone’s identifier String
origin x Origin cell’s x coordinate Integer origin y Origin cell’s y coordinate Integer destination zone Destination zone’s identifier String destination x Destination cell’s x coordinate Integer destination y Destination cell’s x coordinate Integer
travel distance Total distance travelled (cells) Floating point travel time Total travel time (iterations) Integer
gross time budget Agent’s gross time budget Integer net time budget Remaining time budget after trip Integer geometric accessibility Agent’s geometric accessibility value Integer cardinal accessibility Agent’s cardinal accessibility value Integer
Copresence
The encounters between agents that took place during the simulation can be exported to the copresence.csv file, detailed in table 4.8.
Table 4.8: Copresence table structure (copresence.csv). Field Description Data type time Iteration of encounter Integer x Cell’s x coordinate Integer y Cell’s y coordinate Integer agent id Agent’s unique identifier Integer agent group Agent group’s identifier Integer other id Second agent’s identifier Integer other group Second agent’s group identifier Integer
4.4
Summary
This chapter has presented the AxS Model, which implements accessibility and segregation measures based on time geographic concepts. The model’s logic, which represents the translation of concepts discussed in the theoretical frame- work, and its implementation was presented and detailed. The next chapter will explore the model’s behaviour and the effects of the parameters introduced here through a series of verification and sensitivity analysis tests.
Chapter 5
Verification and Sensitivity
Analysis
This chapter presents a set of tests aimed at evaluating the AxS model’s behaviour and outputs under different circumstances, covering the steps of verification and sensitivity analysis of the model building process. As discussed in chapter 3, verification can be understood as the process of making sure a model’s imple- mentation matches its design, while sensitivity analysis consists in systematically evaluating the effects of initial conditions and parameters on the model’s outputs.
The analyses discussed in this chapter were carried out in abstract sce- narios representing hypothetical cities. Those scenarios were designed to isolate the factors influencing the simulation and facilitate the analysis of the model’s outcomes, by allowing each model’s aspect to be analysed separately. The ob- jective of the analysis is to provide a better understanding on how the model works and how the macro-scale results relate to agents’ individual behaviour and to environmental conditions.
In what follows, the verification and sensitivity tests are presented in three sections. Section 5.1 focuses on agents’ navigation process and resulting aggregated flow patterns, section 5.2 focuses on the model’s accessibility outputs, and section 5.3 focuses the model’s segregation outputs.
5.1
Agents’ Navigation and Flow Patterns
The objective of the analyses discussed in this section is to evaluate the agents’ navigation algorithm and resulting movement patterns. The basic abstract en-
vironment used in this set of tests is presented in figure 5.1. It consists of a circular urban area, represented by the light brown cells in the figure, served by a main road network represented by the orange cells. The main road network is structured by two primary roads, one horizontal and one vertical, crossing at the centre of the grid, complemented by a ring road around the city centre and a regular grid spanning the entire urban area. Agents can also move outside the main road network, on regular urban cells, assuming those cells are served by a local road network.
Figure 5.1: Basic abstract environment used in the agents’ navigation sensitivity analysis.
Five exercises aimed at testing the effects of the parameters and en- vironmental factors that influence agents’ navigation were carried out, and are discussed in the following sections. Section 5.1.1 explores the effect of the road network on agents’ movement patterns. Section 5.1.2 focuses on agent’s environ- mental perception. Section 5.1.3 explores how the model’s stochasticity manifests during agents’ decision making process. Section 5.1.4 tests how agents manage to avoid obstacles in the model. Section 5.1.5 tests the effects of the number of active agents in the simulation on the model’s results. Finally, section 5.1.6 summarises the results.