Font Size: a A A

Hot Topic Prediction Analysis Based On P2P Lending

Posted on:2019-05-22Degree:MasterType:Thesis
Country:ChinaCandidate:W W ChengFull Text:PDF
GTID:2359330542455575Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
As the internet finance industry becomes more and more prosperous,People-to-People Lending or Peer-to-Peer Lending(P2P Lending)is also flourishing.P2 P Lending third-party institutions under the Lending community section is that P2 P Lending investors invested platform for one of the main sites about Lending information according to the situation.The hot mining of the corresponding P2 P Lending topics can promptly grasp the focus of investors and the dynamics of P2 P Lending,but also help the relevant government departments to grasp the emergencies in time;and forecasting the popularity of P2 P Lending can play a guiding role in P2 P Lending supervision.In order to find the hot topics that investors are concerned about timely and accurately,this thesis uses the twice clustering method to discover the topic clusters of P2 P Lending and then find the hot topics according to the topic heat calculation formula.Through the first clustering algorithm—ISODATA clustering algorithm to exposure data of bars clustering monthly,and then come to a monthly topic clusters.After that,the number of topic clusters in each month is accumulated and the average number of topic clusters in each month within one year is obtained,and this is used as the initial cluster center of the second clustering algorithm K-means,finally reaching one year topic clusters.This method solves the problem of the deviation about the clustering result caused by the traditional clustering algorithm that uses the given clustering value K simply based on experience.Finally,through the P2 P Lending heat topic formula to get hot topics.In order to predict the topic heat of P2 P Lending,we propose a new topic thermal prediction algorithm based on KNN in this thesis.At the beginning,the KNN algorithm is used to find old topics.And click on a topic that is similar in content to the new topic.The combination of the number about time series and the number of commenting time series predicts the popularity of new topics and trends in P2 P Lending.The final experimental results show that the content of similar P2 P Lending topics in the early stages of the topic which has a similar heat and development trends.
Keywords/Search Tags:P2P Lending, Twice clustering, Heat topic, Topic heat prediction, Topic similarity
PDF Full Text Request
Related items