Prediction of peak occurrence of Dendrolimus punctatus larvae based on Bayes discriminant method

To improve the accuracy of forecasting the peak occurrence of Dendrolimus punctatus Walker, we here used the Bayes discriminant analysis to predict this peak occurrence for the first and second generation of Dendrolimus punctatus larvae based on these data from 1983 to 2016 in Qianshan County, Anhui Province. Our present results showed that this discriminant equation for the first generation was as follows: f⁽¹⁾ = −3.2588‐6.2700x₁ + 1.2870x₂ + 0.7920x₃ + 0.4152x₄; f⁽²⁾ = −14.5215‐8.5710x₁ + 2.9790x₂ + 2.0280x₃ + 0.5031x₄; f⁽³⁾ = −3.5264; f⁽⁴⁾ = −66.8312‐12.5216x₁ + 5.1740x₂ + 4.7162x₃ + 0.6033x₄. And that the prediction accuracy for the first generation was 97.22%. Whilst this discriminant equation for the second generation was as follows: f⁽¹⁾ = −3.536‐1.192x₅ + 1.338x₆ + 0.638x₇₋0.025x₈; f⁽²⁾ = −7.317‐1.337x₅ + 4.240x₆ + 1.010x₇₋0.295x₈; f⁽³⁾ = −16.488‐3.192x₅ + 4.955x₆ + 1.900x₇–0.411x₈; f⁽⁴⁾ = −34.502‐4.184x₅ + 7.484x₆ + 2.583x₇–0.443x₈. The prediction accuracy for the second generation was 85.71%. Overall, our findings revealed that the Bayes discriminant analysis could screen out key factors to significantly improve the prediction accuracy of peak occurrence of Dendrolimus punctatus larvae.