SSD Model Selection Method Based on Machine Learning Algorithm

Abstract The chemical pollutants produced by human activities often lead to the risk of ecological species diversity destruction by affecting the physiological activities of organisms in nature. Ecological risk assessment (ERA) can be used to evaluate the degree of adverse effects of external factors on the ecological environment, provide the basis for taking effective ecological protection measures and formulating reasonable environmental policies, and also an important means to understand the possible adverse effects of ecological health and pollutants on the ecological environment. The species sensitivity distribution (SSD) method is a widely used evaluation method, and its core step is to select the appropriate species toxicity data for curve fitting. In this paper, the basic concept of ecological risk assessment is briefly introduced, and the basic principle and implementation steps of species sensitivity distribution method are described in detail. Aiming at the problem of SSD model selection in water environment, machine learning algorithm is introduced, and corresponding neural network is constructed. Taking the root mean square (RMSE) and sum of squared error (SSE) were as the index, the optimal SSD model suitable for water environment is determined, The SSD model selection method based on machine learning algorithm is obtained. At last, the application of machine learning algorithm in SSD model selection is prospected, and the shortcomings of the existing research and the future research direction are pointed out.