System and Method for predicting the winning bid rate of real estate auction through hierarchical analysis and Transfer Learning

The present invention is a system for predicting a winning bid rate of real estate auctions using hierarchical analysis and transfer learning. For a system for predicting a winning bid rate of real estate auctions which has a control server (10) and a real estate auction article database (20) connec...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: KIM JEONG HUN, RYOU HO SUN, PARK TAE HYUN, JEONG SEUNG HWAN, KIM SANG HOE, JOO SUNG JUN, LEE SANG GYU, KANG EUN, KANG DYOUNG MO, PARK HA EIL, OH KYONG JOO, LEE JUNG TAEK, CHOI YEON JI, LEE SANG HYUN
Format: Patent
Sprache:eng ; kor
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The present invention is a system for predicting a winning bid rate of real estate auctions using hierarchical analysis and transfer learning. For a system for predicting a winning bid rate of real estate auctions which has a control server (10) and a real estate auction article database (20) connected via a network and conducts deep learning on auction articles by operation processing means comprising a computer to predict winning bid rates, the control server (10) comprises: a data processing and variable generating unit (100) which processes auction article data and generates variables; a data classifying unit (200) which classifies the processed auction article data according to regions and usages; a regional stratification classifying unit (300) which classifies the classified auction article data into multiple regional classes based on broadness and narrowness of regions; and a winning bid rate predicting unit (400) which conducts deep learning and transfer learning on the regional stratified auction article data to generate a learning model, and predicts a winning bid rate of an auction article. 본 발명은 제어서버(10) 및 부동산 경매물건 데이터베이스(20)가 네트워크로 연결되고, 컴퓨터를 포함하는 연산처리수단에 의해 경매물건 데이터를 딥러닝하여 낙찰가율을 예측하도록 수행되는 부동산 경매 낙찰가율 예측시스템으로서, 상기 제어서버(10)는 경매물건 데이터를 가공하고, 변수를 생성하는 데이터가공 및 변수생성부(100); 가공된 경매물건 데이터를 지역별 및 용도별로 분류하는 데이터 분류부(200); 분류된 경매물건 데이터를 지역의 광협을 기준으로 복수개의 지역계층으로 분류하는 지역계층화 분류부(300); 및 지역계층화된 경매물건 데이터를 딥러닝 및 전이학습하여 학습모델을 생성하고 경매물건의 낙찰가율을 예측하는 낙찰가율 예측부(400)를 포함하는 것을 특징으로 하는 계층적 분석 및 전이학습을 이용한 부동산 경매 낙찰가율 예측시스템이다.