In researching perceived environment and neighbourhood satisfaction, the main challenge in datasets can be data uncertainty. We use fuzzy logic to deal with this. One of the main advantages of fuzzy logic is non-sensitivity to uncertain and noisy data. The focus of this study was the problem of handling uncertainty when rating scores, this was accomplished by ‘fuzzifying’ the rating values.
For the attention of: Those advising Mayors of urban districts and Mayor’s executive assistants; urban researchers
The problem: In researching perceived environment and neighbourhood satisfaction we faced a great deal of uncertainty and impreciseness in the values provided by our respondents. In other words, our data were not precise enough to be relied on for mathematical prediction models.
On the other hand, subjective ordered variables, e.g. satisfaction, could not be interpreted meaningfully through normal means (multivariate logit models) or even other approaches used in such cases. Hence, we faced blocks and limitations to use of the usual statistical analysis (regression models).
What we did and why: We developed a set of fuzzy estimation systems, each of which predicted the value of neighborhood satisfaction using the values of just one specific subset of the influencing independent variables. The main advantage of fuzzy logic is its non-sensitivity to uncertainty and ‘noise’ in the data. In the proposed model, we did not preserve the original data features. However, we set up fuzzy rules, whereby all necessary antecedents and features, fuzzy sets, and their frequency in the rules were clear.
What our study adds: Any analysis of our model (e.g. sensitive analysis or extracting adversarial examples) will be easy. Experts can insert, remove or edit any rule in our model to adapt, what is in essence, a pretrained system. One of the main advantages of the model we have introduced is its non-sensitivity to uncertainty and noise in the data.
Implications for city policy and practice: We present important implications for policy-makers in our findings and our method. Perceived security, as the most important requirement for achieving a satisfactory neighborhood, must be given special focus through community design policies and guidelines. We also found that when working on subjective data in decision making, it is necessary to use fuzzy modeling approach to handle the uncertainty and impreciseness in the gathered data.
Full research article: Perceived environment and neighbourhood satisfaction: introducing a fuzzy modeling approach by Seyed Mojtaba Fakhrahmad, Seyed Mostafa Fakhrahmad, Ali Soltani & Khalil Hajipour.
