I am working on Parking demand modelling for local area plan. I am into research on fuzzy based parking demand model incorporating human behavior for a realistic prediction. Need your views and comments.
Well, I will keep it high level. To make the model, you will need some pre-steps. I assume that you have already checked theories and you want to base your pattern on a diffusion model (you can also base it on empirical models if you have sufficient data). Next step, I recommend you to follow Takagi-Sugeno`s approach for fuzzy if-then rule-base models since this approach is very close to practical problems. For methodology, you will need an approach for clustering and for clusters, you need some characteristics and so on. I hope I could give you some ideas.
How important is the "fuzzy" aspect for you? I like the previous suggestions if you HAVE to do some fuzzy logic. On the other hand I would be careful that some fuzzy models are pretty uninteresting (like say taking low, medium and high values and simply converting them to a crisp number by division). it is important to do a good model of the process not simply use the tool that appears to be available.
If your real idea is not so much fuzzy but rather probabilistic, I think you could be on a much better track -- for example Ben Akiva and Lerman, or Domencich and McFadden have done great work on building in random/probabilistic aspects to discrete choice models. So for parking, I could see that the choices would be driven by a deterministic (say price) component, but also intangible aspects that are distributed (like say concern for safety and convenience)
See here for a copy of one of these books
https://eml.berkeley.edu/~mcfadden/travel.html
Parking is mentioned in the index on page 146; park-ride, 133
first you need some input variables like: which is closest parking place to youre location, what day is it(working day takes less chance to find parking spot), how many parking spots are in that parking place, etc. After you colected all info now you need to combine them into if-then rules to make membership functions for particular parking places and, if want to go more in detail, for particular parking spot. try taking infos from observing parking places and spots in them, pay attention to preferable spots and frequency of cars.
For Indian condition I feel instead of modelling the parking demand it is better to model the paid parking policy response of the people for the specified parking policy. One of my paper on application of fuzzy logic to model the response to paid parking policy for CBD area of Vadodara city has been accepted for the International Conference organised by Indian Institute of Technology, Bombay in which I took out put as the response of the people for paid parking policy and the input variables were socio economic, travel and parking attributes influencing the response of people.
Other important thing that I would like to tell to Mauro sir and Morton sir is that problems of the parking in Indian condition is very worst and difficult to model.There are no demarcation on the road,no bays for parking and people park here in the haphazard manner. The spaces are found to have high occupancy and low turnover.The only solution that we find is the parking policy for the problem.We are not getting any other direction for how to work in these direction.
Hence, hereby I am attaching one of the conference paper on parking demand assessment for the CBD area of Vadodara city. Plzz refer it and give me fruitful suggestions.
Conference Paper Onstreet parking demand assesment for CBD area of Vadodara city