基于机器学习的众包业务动态定价

Dynamic Pricing of Crowdsourcing Business Based on Machine Learning

提出综合利用K-Means算法、前馈神经网络等机器学习方法进行数据挖掘的动态定价方法。以饷拍平台为例,发现该众包平台的任务地点呈明显的地域簇群现象,合理价格的内在逻辑均隐藏在地理位置信息中。为此,建立了基于地理位置的定价模型,经MATLAB训练后,该模型的损失率、可扩展性及验证结果均表现良好。之后,克服了传统静态模型数据要求高、参数长滞后的缺点,提出了将机器学习与商业数据结合的研究方法。

This paper proposes a dynamic pricing method for data mining by using K-Means algorithm,feedforward neural network and other machine learning methods.Taking the Xiangpai as an example,this paper finds that the mission location of the crowdsourcing platform presents obvious regional clustering phenomenon and the inherent logic of reasonable price is hidden in the geographical location information.Therefore,it establishes a location-based pricing model.After MATLAB training,the model's loss rate,scalability and verification results all performes well.Then it overcomes the shortcomings of high data requirements and long parameters of traditional static model data and proposes a research method that combines machine learning and business data.