Presenting a Spatial Model For Real Estate Appraisal Based On Data-Driven Multi-Criteria Decision-Making Methods

Using spatial data mining and weighted comparison methods can have an efficient role in determining the value of local utilites for efficient indices in real estate pricing. In this study, the factors affecting the price of real estate are divided into four general categories: economic and market factors, physical and welfare factors, neighborhood and access characteristics and organizational characteristics. Organizational characteristics considered as a group of influencing indices in determining property prices that its information is collected from relevant agencies and organizations can contribute significantly in pricing, marketing and management to the different users involved in the real estate sector including buyers, sellers and real estate agencies. The classification proposed assuming stable economic and market factors at the time of user requests and the influence of each index on the quantity and quality is finally divided into three groups that are: Boolean, Multi-Value and Fuzzy/Logarithmic. Where Boolean indices denote those that exist or does not exist on a property that value in order 1 and 0. Multi value indices are those that get more than one value and their influence coefficients are calculated like Boolean indices. Finally, Fuzzy/Logarithmic indices are those that their influence functions are continuous and like fuzzy or logarithmic functions and should be determined by available information and related techniques. Influence coefficient is the influence that the presence or absence of an index affects the property price. For example how presence of elevator affects the property price. Influence function for Fuzzy/Logarithmic indices is the influence that the amount of increasing or decreasing of an index affects the property price. For example how far to main road affects the property price. In this study spatial data mining and compared weighted coefficients are used interactively for gathered information in a crowd-sourced environment. With these methods, indices and impact factors and impact functions are obtained from using different information classifications that have considerable influences in determining local utilities in property pricing. The study area of this research is part of metropolitan area of Kerman (Iran) that is located within 57˚01 ́and 57˚04 ́of longitude and 30˚16 and 30˚19 of latitude from UTM coordinate system(zone 40). Required information includes both spatial and descriptive information. Spatial information includes the location of properties, public places (such as educational, religious, shopping and industrial centers, parks, offices and official buildings) and main and secondary streets. Hence, by using available maps such as maps provided by the National Cartographic Centre (NCC) in 2003 and cadastre maps from registration organizations under the preprocessing GIS ready, the shape files from properties, public places and streets are prepared. All these files are prepared in UTM coordinate system from reference elliptical WGS84. Thus 18053 numbers of properties, 260 numbers of public places and 1009 numbers of streets include main and secondary streets are prepared after preprocessing GIS ready in polygon format to enter to database. Also the description information of 150 properties were gathered from the agencies in the study area.
The results obtained for 16 Boolean indices show that the maximum influence and utility is belong to indices parking, store, elevator and pool and indices the basement and furnished have the lowest utility. Substructure area index is highly dependent on the number of bedrooms. This means that an increase in the number of bedrooms will not necessarily led to an increase in the property price. Therefore the best number of bedrooms up to substructure area 90 squared meter is 2, up to area 120 squared meter is 3, up to area 150 squared meter is 4 and up to area 200 squared meter is 5.
For Multi-Value indices the 15 indices are studied. For the index the number of stories with an increase in class between 2 and 3 percent is added to the price of each property. The index the number of units in a class is inversely related to the property price. This means that if the number of units in each class is higher in the same conditions like area the property price will be lower. For the index the registration document of property the highest utility belongs to full document and the lowest belongs to full devoted document (The full document means the document that completely belongs to its owner with the entire legal rights of a property. When some restrictions apply to the property by the previous owner for example in application style the degree of its ownership will be decreased. The level of restriction in any stage is one sixth of ownership. The restrictions start from full to full devoted document that has the highest restriction.) The registration range of property is a proper index for the regional utilities that depends on different factors. The results show that the highest utility belongs to the registration tag 5 and the lowest belongs to 2773. The organizational restrictions index such as being located in the heritage policy and being located in the urban future plan have a negative influence in local utilities and property price. The results show that being located in the heritage policy has the lowest utility and highest negative influence in property price. For the index view of property there is no utility in the study area and therefore this index is considered as ineffective index.
Finally, for the Fuzzy/Logarithmic indices, calculating the influence function or obtaining a clear process for the changes aren’t easy as two previous groups and the influence rate of these indices is different in various distances. The influences of some indices are estimated by one influence function that are: land area, construction time, land price, reconstruction time, substructure area, distance from shopping centers and distance from main streets. For the indices distance from health centers, distance from educational centers, distance from administrative centers and distance from city center, the influence functions take various modes in different distances. The highest utility for these indices is situating in the distances that are not too far away or to close so that not to take negative effects of too close nor to disturb the appropriate access. There is neutral utility in some distances. This means that situating properties in these distances hasn’t any significant effect on property price. The influence functions for the indices distance from industrial centers, distance from historical and social centers and distance from religious centers are almost the same. According to the results obtained, Boolean and Multi-Value indices obtained with disciplined and more predictable process than the Fuzzy/Logarithmic indices. Finally for evaluation, the prices that were calculated by the system for 10 numbers of properties compared with their actual prices and their accuracy was calculated.