Determining the weights of market features of real estate in explaining their prices is one of the basic objectives of market analysis, performed as part of the property value estimation process. In practice, property appraisers usually settle for basic methods of determining weights, for example based on the principle of ceteris paribus or on the basis of linear correlation coefficients. The article proposes the use of curvilinear correlation coefficients for this purpose; an attempt of such use was made and the obtained results were compared with the weights determined on the basis of linear correlations. The conducted analyses proved that the inclusion of curvilinear correlations at the stage of market analysis, allows for extracting a greater number of features recognized as price-creating, i.e. leads to a smaller loss of market information and is a more reliable tool for determining the weights of attributes in price explanation.