Data providers generally offer only trade estimates, or market data, that covers a small portion of the global supply and demand patterns. These reports are impossible to verify because the analysts do not provide their methodologies.
Farmioc Forecast Models are machine-learning based models of end-of-season yields, for selected crops. We combine several algorithms in a multilevel model, and use understanding of physiological processes in temporal feature selection to achieve high precision in our intraseasonal forecasts, including in very anomalous seasons.
With these Forecast Models, Farmioc provides a comprehensive outlook for global agriculture. Satellite-generated knowledge offered insights into crop conditions, and various agriculture data sets available in our data were used to better inform cumulative year-over-year comparisons.
Source data might be inadequate or delayed for reasons, and Farmioc is developing ways of accurately predicting future yield with minimal data input. Using just NDVI - a measure of plant biomass and therefore crop health - and historical yield values.
We are constantly building more predictive models, which are available to our users. We strongly encourage all those interested in the predictive modeling to pay attention to our forecasts and contact us with any thoughts or questions.