Fractal projection pursuit classification model applied to geochemical survey data

Abstract

A new hybrid exploratory data analysis method, fractal projection pursuit classification (FPPC) model, is developed on the basis of the projection pursuit (PP) and fractal models. In this model, objective classification results are obtained by applying the projection index on the basis of the number-size fractal model. The real-coded acceleration genetic algorithm (RAGA) is used to optimize the projection index to establish the optimum projection direction in the model. Stream sedimentary geochemical data, Gejiu Mining District, Yunnan Province, China, were chosen in a case study to demonstrate the processing data analysis using FPPC. The results show that the anomalies are associated with known mineral deposits in the eastern part of the Gejiu District, and correlated with faults and granite in the western part of the study area. It is demonstrated that FPPC can be a powerful tool for multi-factor classification analysis and provide an effective approach to identify anomalies for mineral exploration. ► A new hybrid model that fractal projection pursuit classification model is developed. ► Real-coded acceleration genetic algorithm is used to optimize the projection index. ► Projection index is applied for objective classification based on the number-size fractal model. ► FPPC is demonstrated to be an efficient tool for processing geochemical data.