IDP: An Intelligent Data Prediction Scheme Based on Big Data and Smart Service for Soil Heavy Metal Content Prediction

In the application of regression prediction through big data technology, error between predicted value and true is often large. order to reduce prediction, this paper proposes an Intelligent Data Prediction (IDP) scheme for Smart Service. It uses Least Squares Support Vector Machine (LSSVM) as basic model. Since there no standard procedure determining main parameters LSSVM, improved Particle Swarm Optimization (MBPSO) algorithm used simultaneously optimize LSSVM. The disadvantage PSO precocity due disappearance population diversity. Based on this, Improvement strategy MBPSO aims continuously generate “More” “Better” particles. First, in avoid early particle diversity, re-adjusted inertia weight learning factor. Secondly, a renewable access proposed allow part disappeared regenerate. Finally, method global optimal adjustment introduced help particles find flight direction. verify effectiveness MBPSO, 9 test functions are performance. results show that MBPSO’s optimization speed, best mean all perform best. Taking farmland soil heavy metal sets Dongxihu District Hannan Wuhan City examples application, content metals Cr Pb was predicted. IDP closer actual value, three index values significantly lower than other models. Especially content, compared with LSSVM model, errors two regions reduced by 25.67% 20.70% respectively. We can conclude has practical significance prediction.