Modelling of leaf folder populations (Cnaphalocrocis medinalis) in Paddy: A count time series approach

Abstract

The present study was conducted to model the Leaf Folder pest population in Paddy at Agricultural Research Station, Bapatla. The secondary data between 2012-13 to 2020-21 was considered based on data availability. Correlation and stepwise regression were used to check the relationship between pest and weather parameters. The Minimum temperature had significant negative correlation and maximum temperature and maximum temperature and rainfall were significantly contributed, and having negative impact on Leaf Folder population. Count time series and machine learning models are used for fitting the Leaf Folder dataset. INGARCH-ANN model outperformed well than INGARCH, ZIPAR, ZINBAR, ANN models based on error comparison criteria (MSE and RMSE) and the statistical significance between the models utilized in the study were determined by Diebold- Marino test statistic (DM test). The order of prediction accuracy of the models under consideration is INGARCH-ANN>ANN >ZINBAR>ZIPAR>INGARCH.