Compliance model dairy animal welfare

Netherlands Food and Consumer Product Safety Authority (NVWA)

The NVWA checks whether dairy farmers take adequate care of their animals. They must comply with animal welfare rules. This algorithm helps companies find where the risks are highest.

Last change on 22nd of May 2026, at 11:05 (CET) | Publication Standard 1.0

• Publication category: Impactful algorithms
• Impact assessment: Field not filled in.
• Status: In use

General information

Theme

Economy

Begin date

2025-03

Contact information

https://www.nvwa.nl/over-de-nvwa/contact

Link to publication website

Nog niet bekend

Responsible use

Goal and impact

This algorithm is designed to make the most useful use of our inspectors' time. Currently, for example, we carry out inspections at

1. Farms selected by chance. This way, we keep a good overview of how dairy farms in general are doing.
2. Farms about which a report has been received, because animals may be treated badly. These reports can be made by citizens, vets or other companies.
3. Farms selected on the basis of risk factors, such as high calf mortality.

Furthermore, we mainly want to inspect farms where it seems more likely that rules are not followed. This algorithm helps find such farms. This is followed by an inspection as we always do. It leads us to do more inspections at companies where something might be wrong, and less at companies where everything is fine. Inspecting where the risks seem greatest is required by law; with this algorithm, we can do it better.

Considerations

This algorithm helps us find farms where they are more likely not to comply with animal welfare rules. When violations occur, we can intervene and make sure the situation improves. That is a big advantage. The algorithm also ensures that we can be sure that we calculate risks in the same way for all farms, with information that we are allowed to use for that purpose. That is also an advantage.

A possible disadvantage is that the algorithm learns from previous inspections. This could lead to the algorithm suggesting the same types of companies each time, or skipping them.

Another possible disadvantage is that inspectors might see more problems if they know they are performing an inspection on a company selected by an algorithm. This could lead to more violations being found on some companies because the algorithm is better at finding them, not because they are worse at complying with the law. But because we know these risks are there, we have also taken measures against this.

We believe that all in all, the advantages outweigh the disadvantages.

Human intervention

For each farm with at least 20 dairy cows, this algorithm predicts the probability of non-compliance with animal welfare rules. Of the farms with the highest probabilities, we make a list to inspect. The number of farms for final inspections is determined by humans, and this list itself is also checked by hand. If we discover possible errors or improvements, we adjust the algorithm or correct the list of companies. Then each company on the list is visited by an inspector. The algorithm only gives advice on which dairy farms to inspect. The algorithm does not make decisions or give advice on the inspection itself.

Risk management

As we wrote earlier, we see two main risks in using this algorithm. The first is that we might inspect some types of dairy farms too often or too little. To prevent this, we are taking several measures. First, we always continue to do inspections of dairy farms that are selected by chance. This way, we check whether the algorithm really helps detect more problems than we would otherwise. Second, we compare the list of dairy farms suggested by the algorithm with the list of all dairy farms out there. If we find that the algorithm no longer predicts well enough where there are problems, or always picks or skips the same type of farms, we improve it, or don't use it. We also ask inspectors to share their experiences based on the inspections performed. Their feedback helps us to further improve the algorithm.

The second risk is that the algorithm could influence what inspectors think of the companies they inspect. To make sure this happens as little as possible, we explain well that the algorithm only calculates a probability. The fact that a dairy farm has been suggested by the algorithm really does not mean there is anything wrong with it; only that it looks like dairy farms where we have found a problem before.

In general, we think the risks of the algorithm are small. There is a lot of human control; and the total number of farms inspected on the advice of the algorithm is not large.

Legal basis

1. The Animals Act
2. Animal Husbandry Decree

Links to legal bases

• The Law on Animals: https://wetten.overheid.nl/BWBR0030250/
• Animal husbandry decree: https://wetten.overheid.nl/BWBR0035217/

Operations

Data

The algorithm uses data from 3 sources:

1. Our own inspection results,
2. the system in which dairy farmers must report the births, movements and deaths of their animals ('I&R'),
3. the 'agricultural census', in which all farmers must report the characteristics of their farms annually

Links to data sources

• I&R: https://www.rvo.nl/onderwerpen/identificatie-en-registratie-dieren/runderen-melden
• Agricultural census: https://www.rvo.nl/onderwerpen/gecombineerde-opgave

Technical design

So far, the algorithm has been revamped with every use. This involves comparing different supervised machine learning techniques and choosing the best predictive one. We use these techniques to automatically learn the relationship between business characteristics and inspection results. To do so, we divide all available inspections into a training set and a test set. The inspections in the training set are used to learn the correlation; the inspections in the test set are used to test whether the computed correlation is also predictive of inspections that the algorithm has not seen before.

External provider

The algorithm was created by the NWVA itself.