The Influence of Chosen Statistical Methods of Detecting Outliers on Property Valuation Result

The subject of the thesis concerns the application of selected statistical methods searching for outliers in the process of determining the value of real estate, based on a functional model adjusted to market data.

The collected research material consisted of data on land properties, which were the subject of transactions on local markets, for which there was no information regarding the specific conditions of concluding the sale agreement. After the initial selection of data regarding the purpose of the property in the local plan, the type of property rights being sold and the size of the shares sold - a functional model was adjusted to the obtained data, showing the relationship between the price being the dependent variable and the features of the property being the independent variables. Then, two statistical methods of searching for outliers which are significantly different in their algorithms, i.e. Cook’s distance and robust estimation method called Pope’s method, were applied to each model. The last stage was to determine the model values of selected properties and to compare the obtained results with the known transaction prices of the parcels being the subject of the valuation.

The conducted research allowed for the verification of the influence of significantly different statistical methods searching for outliers on the property valuation result and its accuracy.