Harvest Forecast Analytics helps you understand how Source’s harvest predictions are created and what factors influence their accuracy. By visualizing key input parameters and model outputs, it enables you to identify trends, explain deviations, and build confidence in your forecast data.
What it does
The tool gives a transparent view of how your predicted harvest is formed. It shows the relationship between your inputs, Source’s model forecasts, and realized results from your greenhouse.
With these insights, you can:
Verify the accuracy of your forecasts
Spot unusual or inconsistent data
Communicate confidently with your sales team about expected harvests
Understand how factors like fruit weight, plant load, and climate conditions influence yield
Before getting started
For the analytics to provide meaningful insights, make sure that:
You’ve already turned Source’s forecast into your own harvest commitment (this provides the historical data needed for comparison)
You’ve been using Harvest Forecast for at least eight weeks
How it works
Harvest Forecast Analytics combines multiple data sources to explain how each forecast was built. It includes two main views:
Chart view
The chart shows visual trends across the current week and the eight weeks before it. By default, it displays your Predicted, Committed, and Realized harvest volumes (in kilograms).
You can add additional input parameters to the chart, such as fruit weight, waste, or daily radiation.
Each parameter includes two lines:
Target values – the inputs you entered
Realized values – the actual outcomes recorded later
Selecting a date in the chart defines the “cutoff” day. This determines which forecasts and input data are shown in the table view below.
Table view
The table presents the same data in a detailed, week-by-week format. Each column represents a week, and each row shows a specific metric.
By default, each week’s data uses the last day of that week as its cutoff point. This means the values shown for week 40, for example, represent the forecast as it was known at the end of that week.
You can change this by selecting a different day in the chart. For instance, choosing Monday instead of Sunday if you want to see the forecast as it looked when you sent your harvest commitments.
The table allows you to trace how predictions evolved over time and compare them with realized results once the week is complete.
Interpreting the data
When you select a week to analyze, the product displays how the forecast for that week was built using data from the previous eight weeks.
Each section represents a key driver of your harvest forecast:
Harvest
Predicted net – Source’s forecasted harvest weight
Committed – Your sales commitment for that week
Waste – The expected or realized percentage of discarded fruits
Plant
Fruit weight – The forecasted or committed average fruit weight
Harvestable fruits – The total number of fruits predicted to be ready for harvest
Strategy
Color scale – The ripeness level you’ve planned for harvest (lower values mean greener fruits)
Rounds – The percentage of greenhouse paths planned for harvest
Climate
Daily radiation (8-week average) – The combined realized and forecasted light levels
Temperature (8-week average) – The average greenhouse temperature over the same timeframe
As you move closer to a given harvest week, climate and plant parameters become more accurate because more realized data is included in the averages.
Finding insights
Harvest Forecast Analytics can help you identify why a forecast may have been higher or lower than expected.
For example, if a predicted harvest was overestimated, you can explore whether this was caused by a lower realized fruit weight, fewer harvestable fruits, or unexpected waste.
Some ways to use this tool:
Validate forecast accuracy: Compare predicted vs realized harvests to evaluate model performance.
Investigate anomalies: Check if sudden changes in fruit weight or plant load explain unexpected results.
Support communication: Use the data to clarify differences between forecasts and actual outcomes during discussions.
Frequently asked questions
Can I export the data?
Export functionality isn’t available yet, but it’s planned for a future release.
Why does the realized harvest line look the same across all weeks?
The realized line represents the actual harvest for the selected week. It stays constant across the chart so you can easily compare it with predicted and committed values.
Summary
Harvest Forecast Analytics gives you full visibility into how forecasts are created, helping you understand your greenhouse data in context.
By analyzing patterns and comparing targets with outcomes, you can make more informed cultivation decisions and improve the accuracy of future forecasts.