Cost Relationships In Housing Physically-Handicapped People

This thesis describes a fresh approach to regression modelling in the construction field - in which building plans are generated specially for the purpose. Generating the plans in the course of the modelling is necessary because information from other sources is not sufficient, in quantity or quality. The work follows an earlier study of regression analysis in the context of hospitals. The context of this thesis is -housing for physically-handicapped people. Housing for physically-handicapped people is chosen for two reasons. The primary reason is that a context which is small-scale is necessary, since generating plans is entailed. Secondarily, research to determine how best to make provision for physically-handicapped people in housing is required in its own right. There is a view underlying the research that regression modelling is about learning. This view holds from the first thought about the population through to the use of the model for the prediction of prices. The modelling provides a vehicle for organised investigation and recollection. The thesis comprises three parts and conclusions. Part I sets the scene of the research; Part 2 deals with Vi generating the house plans; and Part 3 examines the method of modelling and what was obtained from it. Part One of the thesis gives an explanation of the customary approach to the regression modelling of construction prices and the problems that this approach poses. Plans generated specially for the purpose of the modelling can overcome the problems. The generation of plans is a problem in itself, however, in effort and time. Here an automated method of generating the plans would help. This is only possible for small-scale buildings at present. Part Two of the thesis explains the generating of the house plans. The house plans may be generated by adjusting other house plans or they may be generated from scratch. An example is provided of generating the house plans as an adjustment while a step is taken towards generating them from scratch by computer. In the latter context, an existing sample of house plans is analysed into a set of statements or commands. This goes towards the beginning of a library of statements for use in the course of generating handicap house plans from scratch by computer. Part Three of the thesis is a description and appreciation of the information, including the models, that is forthcoming in the course of the modelling. A range of models produced in the course of the research is shewn. Data are generated to meet requirements as needed. vii Some of the models are at great detail. Their purpose is to explore and map out relationships and then fall back to simpler models as predictors. The place of the simple models is for use as predictors at am early stage in planning the development of buildings - when information is sparse. So models that involve few variables are needed. The conclusion of the thesis is that price models obtained from specially-generated plans are more satisfactory than those normally attainable with construction project data; while it appears that generating plans by computer is not an impossibility in the case of straight-forward housing.