基于核加权KNN和多目标优化的众包平台定价系统设计

Pricing System Design of Crowdsourcing Platform Based on Kernel Weighted KNN and Multi-objective Optimization

随着科技的发展,众包模式越来越受到人们欢迎,而任务定价合理程度关乎众包平台的发展。因此,本文针对任务定价问题,结合数据挖掘和多目标优化理论,设计了基于核加权KNN和多目标优化的众包平台定价系统。首先,分析影响任务定价的相关指标,构建基于核加权KNN的接单判别模型,并利用PSO算法优化参数,增加模型预测精度与鲁棒性。其次,以任务完成率最大、任务总价最小为目标函数,构建多目标优化模型。通过贪心法得出每增加1%任务完成率所需要最少的加价额度,采用有序样品最优聚类,根据最优分割点确定最佳定价方案,保证众包平台能在最小的代价下得到最好的发展。最后,提供众包平台定价系统的设计建议与改进方向。

With development of science and technology,crowdsourcing mode is more and more popular,and reasonable degree of task pricing is related to development of crowdsourcing platform.Therefore,the article designs a crowdsourcing platform pricing system based on kernel weighted KNN and multi-objective optimization aiming at problems of task pricing,and combining data mining and multi-objective optimization theory.First of all,to analyze related indicators affecting task pricing,construct order discriminant model based on kernel weighted KNN,and optimize parameters with PSO algorithm to increase accuracy and robustness of the model.Secondly,to construct multi-objective optimization model with objective function of maximum task completion rate and minimum total task cost.With greedy method,to obtaine minimum price adding amount for 1%increase each of task completion rate,to determine the best pricing scheme according to optimal segmentation point,with ordinal samples optimal clustering to ensure the best development of crowdsourcing platform at the minimum cost.Finally,to provide design proposal and improvement direction of pricing system of crowdsourcing platform.