Blueberry farm

How Our Customers Use Our Forecasting Tools

We wanted to highlight some of the different ways Bitwise Ag customers use our GreenView forecasting tools.

The reality of modern farming means growers need to understand human factors, market influences, and logistic behaviours in order to accurately forecast. The example we’re showing today highlights the discrepancy between forecasted blueberry picking amounts and actual picks due to human decision-making in the field.

GreenView forecasting tools

What you can see above is the summary page our customers see, which includes historical data on different blocks that were picked. At the top you can see the different blocks, dates for each week, and the GreenView estimate about what’s going to be picked each week for those blocks.

The middle section shows the actuals – the data the customer gave us on what was actually picked on a certain week. Testing involved comparing the forecasted blueberry picking amounts with the actual picks provided by the customer.

And at the bottom, in the colourful section, we show how close our estimates were to the actuals. On a weekly basis, we have a target error rate under 15% and for seasonal forecasts we like to get within 5%, which you can see we’re achieving here.

This blueberry customer is really interesting because GreenView saw the trend of the fruit getting lighter over an 8 week period. We started with 16 tonnes available for picking at the beginning of March and by mid April there were 12 tonnes, so decreasing about a tonne every 2 weeks. Which is pretty standard and what we’d expect to see.

But then we got the data from the customer and we could see it was very different. The first week in block 3 they picked 11.8 tonnes, and when they were picking 2 weeks later there were 21.8 tonnes available! The reason for this is that some of the fruit was left on the plant from the previous pick.

Once we identified these discrepancies, we discussed them with the customer to understand the human factors influencing the picking decisions. The customer said that the GreenView forecasts are bang on, but their strategy was to alternate heavy and light picking weeks based on market conditions, and we took that into account for analysis.

The customer uses the GreenView forecasts to assess what they’ve got to pick, and when to go heavy and light. Despite variations in actual picks compared to forecasts, the overall error percentage was low (1.5%) over an eight-week period, demonstrating the effectiveness of the forecasting tool in guiding picking decisions.

This example really shows that forecasting is not black and white – it’s a complicated science and weather impacts, human behaviours, market conditions and logistics all affect what hits the market.

Want to know more? Book a free demo of GreenView or contact us today on