Study of the Prediction of Development Stage and Population Size of Soybean Aphid in Northern Liaoning, China

A model for predicting the days for aphids to reach 10 thousand/100 plants in Northern Liaoling province was established by stepwise regression. In order to include enough parameters in the stepwise regression model to predict the aphid population, parameters were carefully selected by path analysis at first, and then a multiple regression model was established (F=0). At the same time, a stepwise discriminant model was established too. Integrated prediction was conducted by the combination of these two methods. Prediction fed with historical data had 100% fitness. Forecasting tests in the past two years were all correct.