This is the project I did for the Social and Technological Networks course I took as part of my MSc Artificial Intelligence at the University of Edinbrugh. Using data from the Nederlandse Spoorwegen, I modeled which cities are best to commute from if you work near Amsterdam Central Station (ASD) but don't want to pay Amsterdam rent, based on the following factors:
- Crowdedness \( c \): How busy the train is
- Duration \( d \): How much time it takes to get to ASD
- Flexibility \( f \): How many ways there are to get to ASD
- Punctuality \( p \): Whether the train(s) will arrive on time
- Switches \( l \): How many times the commuter has to switch trains
I combined these factors into a single stress measure \( S \) as follows, where \( \alpha \)s are weights for each factor and \( O \) is the set of possible options:
$$ S = \min_{\forall o \in O} \left( \alpha_c c_o + \alpha_d d_o + \alpha_p (1 - p_o) + \alpha_s s_o \right) + \alpha_f (1 - f) $$
By scraping data from the Dutch national railway operator Nationale Spoorwegen (NS), I could then make the following plots of how “stressful” a commute to Amsterdam would be for each train station in the Netherlands, for different \( \alpha \) settings.
Stress maps with different settings for \( \alpha \)s; green indicates low stress and red indicates high stress. Amsterdam Central Station (ASD) is marked in black. The mixed model uses my preferences as \( \alpha \) settings.
Using settings from the bottom-right mixed model, the best cities to commute from are mostly in the Randstad and the area north-west of Amsterdam. See Appendix A of the report for a list of the top-thirty stations according to this metric.
More info: