关键词: 访问数据预测, K-means算法, BG/NBD模型, 协同过滤推荐, 数据预存储

Abstract: To save the battery and relevant energy-efficient energy of smartphones,a data forecasting-based strategy for smartphone energy-efficiency based on predictability of users accessing data was proposed.The known users were classified into clusters with mixed variable attributes-based improved K-means algorithm,and the similar user group was established.The expectation of user's continuous search was predicted by using BG/NBD model.Aiming at the valuable users,combined with collaborative filtering recommendation,the data to be accessed by user was pre-fetched based on historical information of similar users.The data pre-storage mechanism was utilized to pre-store the above predicted data in smartphones to achieve the goal of energy saving in the way of reducing the communication times.Compared with other methods,the experimental result showed that the strategy could reduce the energy consumption over 13% on the condition of meeting user satisfaction.

Key words: data forecasting, K-means algorithm, BG/NBD model, collaborative filtering recommendation, data pre-storage mechanism