Review article

The article discusses approaches to forecasting thecost and timing of implementing residential real estatedevelopment projects using regression modeling. Basedon empirical data on 40 projects of multi-apartment residential buildings in St. Petersburg, there has beendeveloped and tested a model for estimating the unitcost of facility construction. The construction andtechnical and economic parameters of the buildingswere used as factors: building volume, building area, number of floors, volume of the underground part. The use of multiple linear regression allowed obtaining high values of the accuracy criteria of the model ( R ²= 0,85, MAPE=1,75 %), which makes the developed dependencies applicable for estimating the costs of projects at the pre-investment stage. The results obtained were compared with results inforeign studies based onartificial intelligence methods (ANN, LASSO, Random Forest).It is shown that with a limited amount of data, regression models provide comparable accuracy while maintaining interpretability and analytical transparency. The practical significance of the research lies in the possibility of using the developed models when forming the budget for housing construction projects.