Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production
This study develops interrupted models to forecast Indian food grain production under the influence of an intervention event. A Pettitt test identified 1987 as a significant change point (p < 0.0001), indicating a structural shift in the agricultural production system. The study employs interrupted...
| Main Authors: | , , |
|---|---|
| Format: | Journal Article |
| Language: | Inglés |
| Published: |
Springer
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10568/174392 |
| _version_ | 1855543012270014464 |
|---|---|
| author | Chitikela, Gayathri Rathod, Santosha Vijayakumar, S. |
| author_browse | Chitikela, Gayathri Rathod, Santosha Vijayakumar, S. |
| author_facet | Chitikela, Gayathri Rathod, Santosha Vijayakumar, S. |
| author_sort | Chitikela, Gayathri |
| collection | Repository of Agricultural Research Outputs (CGSpace) |
| description | This study develops interrupted models to forecast Indian food grain production under the influence of an intervention event. A Pettitt test identified 1987 as a significant change point (p < 0.0001), indicating a structural shift in the agricultural production system. The study employs interrupted time series and machine learning models, including autoregressive integrated moving average (ARIMA), interrupted ARIMA, artificial neural network (ANN), and interrupted ANN, to forecast key food grains such as rice, wheat, coarse cereals, total cereals, pulses, and total food grains. The intervention effect was significant for most commodities, with 1987 coded as 1 and other years as 0. The interrupted models particularly interrupted ANN, achieved superior accuracy in both training and testing datasets and the highest error reductions, with up to 96.45% for pulses and 99.06% for total food grains in the training dataset and 99.05% for pulses and 91.45% for total food grains in the testing dataset. Despite a negative production effect in 1987, the overall trend remained positive, with average annual production increases of 14.54, 15.28, 3.73, 33.65, 1.63, and 35.18 lakh tons for rice, wheat, coarse cereals, total cereals, pulses, and total food grains, respectively. While interrupted ARIMA models capture linear patterns, they often fail to address nonlinearities. This study addresses this gap by developing interrupted ANN models to capture nonlinearities in Indian food grain production data. This approach offers valuable insights into addressing structural shifts in agricultural systems, improving production forecasting, and supporting evidence-based policy-making in developing economies. |
| format | Journal Article |
| id | CGSpace174392 |
| institution | CGIAR Consortium |
| language | Inglés |
| publishDate | 2024 |
| publishDateRange | 2024 |
| publishDateSort | 2024 |
| publisher | Springer |
| publisherStr | Springer |
| record_format | dspace |
| spelling | CGSpace1743922025-11-12T05:00:20Z Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production Chitikela, Gayathri Rathod, Santosha Vijayakumar, S. rice wheat cereals time series data analysis developing countries farming systems forecasting artificial intelligence This study develops interrupted models to forecast Indian food grain production under the influence of an intervention event. A Pettitt test identified 1987 as a significant change point (p < 0.0001), indicating a structural shift in the agricultural production system. The study employs interrupted time series and machine learning models, including autoregressive integrated moving average (ARIMA), interrupted ARIMA, artificial neural network (ANN), and interrupted ANN, to forecast key food grains such as rice, wheat, coarse cereals, total cereals, pulses, and total food grains. The intervention effect was significant for most commodities, with 1987 coded as 1 and other years as 0. The interrupted models particularly interrupted ANN, achieved superior accuracy in both training and testing datasets and the highest error reductions, with up to 96.45% for pulses and 99.06% for total food grains in the training dataset and 99.05% for pulses and 91.45% for total food grains in the testing dataset. Despite a negative production effect in 1987, the overall trend remained positive, with average annual production increases of 14.54, 15.28, 3.73, 33.65, 1.63, and 35.18 lakh tons for rice, wheat, coarse cereals, total cereals, pulses, and total food grains, respectively. While interrupted ARIMA models capture linear patterns, they often fail to address nonlinearities. This study addresses this gap by developing interrupted ANN models to capture nonlinearities in Indian food grain production data. This approach offers valuable insights into addressing structural shifts in agricultural systems, improving production forecasting, and supporting evidence-based policy-making in developing economies. 2024-03-21 2025-04-30T06:48:27Z 2025-04-30T06:48:27Z Journal Article https://hdl.handle.net/10568/174392 en Open Access application/pdf Springer Chitikela, Gayathri, Santosha Rathod, and S. Vijayakumar. "Change point-driven interrupted time series and machine learning models for forecasting indian food grain production." Discover Food 5, no. 1 (2025): 1-17. |
| spellingShingle | rice wheat cereals time series data analysis developing countries farming systems forecasting artificial intelligence Chitikela, Gayathri Rathod, Santosha Vijayakumar, S. Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production |
| title | Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production |
| title_full | Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production |
| title_fullStr | Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production |
| title_full_unstemmed | Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production |
| title_short | Change point-driven interrupted time series and machine learning models for forecasting Indian food grain production |
| title_sort | change point driven interrupted time series and machine learning models for forecasting indian food grain production |
| topic | rice wheat cereals time series data analysis developing countries farming systems forecasting artificial intelligence |
| url | https://hdl.handle.net/10568/174392 |
| work_keys_str_mv | AT chitikelagayathri changepointdriveninterruptedtimeseriesandmachinelearningmodelsforforecastingindianfoodgrainproduction AT rathodsantosha changepointdriveninterruptedtimeseriesandmachinelearningmodelsforforecastingindianfoodgrainproduction AT vijayakumars changepointdriveninterruptedtimeseriesandmachinelearningmodelsforforecastingindianfoodgrainproduction |