A method for generating a learning model for predicting real estate transaction price

摘要

The method of generating a learning model for predicting real estate transaction price of the present invention comprises the steps of dividing a loaded data set into a training set and a test set based on a specific time point; forming an instance by extracting or adding a feature for each data of the training set; generating a decision tree-based time series trend prediction learning model by performing gradient boosting based on the instances of the training set; forming instances on the test set data by feature extraction or addition in the same manner as the training set; generating virtual data in which the actual transaction price is predicted by inputting the instance characteristics of the test set data to the learning model; Measuring the error between the actual transaction price of the virtual data and the test set data, when the target performance is not met, by varying the hyperparameter of the learning model creation step to further expand the learning tree, and fixing the learning model when the target performance is satisfied includes