Research On The Site Selection Of Urban Hotels Based On Transfer Learning

With the development of smart city concepts and machine learning technology,urban computing has become a popular research area.This paper makes an in-depth study on the development trend of urban hotels in urban computing with the help of machine learning methods,mainly using transfer learning methods to analyze urban data.Compared with the existing work,this paper proposes different implementation schemes in the application of city similarity measurement methods and transfer learning methods in the prediction of urban hotel development trends,and the correctness of the schemes is verified by experiments and actual algorithm visualization system applications and feasibility.The author independently completed the following works:The data in the city has many types and wide distribution.How to extract the data types with high correlation with the target problem is a key problem that needs to be solved in the data preparation step.This paper proposes a city data collection method based on the location and quantity information of city points of interest.Data from 15 different types of cities are collected for subsequent experimental work;The core of transfer learning is to find areas with high similarity and transfer their related knowledge,so the measurement of similarity between areas is a key issue in transfer learning.In particular,for urban data,this paper proposes a city feature extraction and similarity calculation method based on gridded urban data.Experimental results show that the calculation method can correctly measure the similarity between the two cities;Compared with developed cities,the need of urban hotel location planning is often greater in less developed cities,but the amount of data in less developed cities is small,and the low degree of fitting of machine learning algorithms based on a single city is a difficult problem.This paper uses the method of transfer learning and uses multi-source similar city data to train the prediction model of urban hotel development scale.The experiment proves that the algorithm of urban hotel development scale prediction based on transfer learning can effectively improve the accuracy of urban hotel distribution and scale prediction;This paper implements a visual system for the prediction model of city hotel development scale,and displays the actual operation effect of the city hotel development trend prediction algorithm in an intuitive visual way.This visual system has three major functions,they are searching and locating the target city,visually displaying the city grid results,and predicting the distribution scale of city hotels and displaying the results.Experiments prove that the city similarity measurement method and city hotel development trend prediction algorithm proposed in this paper are correct and feasible in practical visualization applications.It is verified by experiments that the urban similarity calculation algorithm and similaritybased transfer learning method proposed in this paper can accurately predict the development trend of urban hotels and give useful suggestions for the site selection decision of commercial hotels.Also it can provide methods for other types of corporations on site selection basis.