01-05 December 2025
INCOIS, Hyderabad, India.
| Abstract Submission No. | ABS-07-0137 |
| Title of Abstract | Forecasting Global Seaweed Production Trends Using Autoregressive Time Series Models: Machine Learning Approach |
| Authors | Digvijay Singh Yadav*, Bhavik Kantilal Bhagiya, Pankaj D. Indurkar, Vaibhav A. Mantri |
| Organisation | CSIR-Central Salt and Marine Chemicals Research Institute |
| Address | Gijubhai Badheka Marg, Takhteshwar Vidyanagar Bhavnagar, Gujarat, India Pincode: 364002 E-mail: digvijaysingh.yadav@gmail.com |
| Country | India |
| Presentation | Oral |
| Abstract | With the global population rising and traditional resources under strain, seaweed aquaculture is emerging as a vital solution for future food security and sustainable development. Data available in literature reflects a significant global shift in seaweed aquaculture, rising from 6.58 million tons in 1992 to over 35.14 million tons in 2022, with market value increasing from USD 3.12 billion to USD 16.8 billion in the same period. Most of this growth (97%) stems from aquaculture rather than wild capture. A strategic production shift from brown to red seaweeds, especially Eucheuma, Gracilaria, and Kappaphycus spp., underlines market-driven demand for hydrocolloids like carrageenan and agar. The present study aims to forecast global seaweed production using classical statistical time series models ARIMA, SARIMA, SARIMAX, and PROPHET based on data sourced from the FAO FishStatJ database (1950⿿2022). The models were developed to predict future trends based on historical data. Model validation has been done by splitting the dataset into two groups, of which 70% was used for model training and 30% for testing. Performance was assessed using Root Mean Square Error (RMSE) and regression coefficient (R²), and the R² values were more than 0.7 for the tests, indicating a strong model fit. Based on these findings, we have predicted the country-wise global production and the choice of seaweed species by year 2030. Results confirmed SARIMAX outperformed ARIMA and SARIMA across most forecasting scenarios, particularly in regions like East and Southeast Asia with strong exogenous drivers. The models outlined a continued upward production trend globally, though regional dynamics like species shifts and socio-environmental constraints introduce nonlinearity. The work highlights the critical role of various forecasting models in marine Agri-economics and lays a foundation for future hybrid approaches incorporating deep learning for robust, long-term predictions. |
| Are you part of IIOE-2 endorsed project | no |
| Keywords | ARIMA, Sustainable Aquaculture, Food Security, SARIMAX, Marine Economics |
| For Awards | yes |
| Date Of Birth | 19-07-1988 |
| ECSN Registration Number | IIOE2-ECSN-0214 |