Now that you know why we began almond forecasting, you might be curious, how do we do it?
Our approach is simple — we analyze data.
First, let’s start with things we don’t do:
- Telephone surveys aka opinions
- Visit orchards (well, not for forecasting purposes…)
- Collect almond samples
- Count nuts (we are nuts, but not THAT nuts)
- Analysis through classical statistics
You might be thinking — no classical statistics, I thought Bountiful was data-driven…We are, and here’s how:
What we do:
We analyze satellite imagery, weather, geographical, and historical data with the below analysis techniques:
- Bayesian data analysis methods
- Machine learning
The datasets we analyze play a role in shaping the outcomes for each growing region. Bayesian data analysis methods coupled with machine learning allows us to continuously learn and adapt our models as we receive new information. This allows us to produce multiple forecasts per crop year. Four to be exact starting in April.
Data allows us to remove bias, give context to yield, and allow us to accurately forecast almond supply 2x better than the industry standard.
Learn more about our technology here https://bit.ly/2XrTPKf