A residential satisfaction decision support system for affordable housing

This research used neural networks to develop a decision support system, and model the relationship between one's living environment and residential satisfaction. Residential satisfaction was investigated at two affordable housing multifamily rental properties located in Atlanta, Georgia. This research used two approaches. The first approach examined the survey data collected from both properties, Defoors Ferry Manor and Moores Mill, separately. The second approach analyzed survey data by consolidating data from both sample populations into one data file. This research analyzed the surveyed data in both approaches using four techniques: frequency distributions, chi-square analysis, correlation analysis, and neural network analysis. In the first approach, the neural network accurately categorized ninety-eight percent of the cases in the training set and ninety-three percent of the cases in the validation test set. The second approach used two trials were used for the neural network. The first trial used one hundred, twenty-five rows to train the data, and the remaining fifty-three rows to validate the network; the neural network correctly categorized eighty-three percent of these cases. The second trial used one hundred, seventy-eight rows to train the data and the same data rows to validate the network; the neural network correctly categorized ninety-three percent of these cases. This research represents a first attempt to use neural networking to model the relationship between one's living environment and residential satisfaction.