top of page

Data collection in Aquaculture – how to manage and what is it good for?

Knowing me, you probably know that I believe data collection is one of the most critical procedures in our industry. Correct and accurate data collection cannot be overstated.

Nonetheless, we must keep in mind that it must be a very easy process, so we must ensure that we have a suitable tool in place.

Data collection will not be accurate or continue if it is not easy to handle. Therefore, one of our goals is to make all processes as simple as possible.

However, let's say we got it right, and now we have the most reliable data available. What can we do with it?

There are a few levels of data usage:

- Simple reporting

- Basic analyzing

- Data mining

- BI tools for a deeper level of analyzing

- Modeling – specific modeling of the performances, based on local data

First thing first – reports: Well-built reports may provide new insights on the farm. I met growers who suddenly discovered, only after a better arrangement of the data, that there is a difference between the feed quantities they thought they served and the real amounts the fish ate, leading them to assume they are losing large quantities of feed without being aware of it. It is possible for a number of reasons, such as a feeding machine not being calibrated or feed being stolen from the storerooms, and more. However, all of them result in significant losses of money.

In other reports you can compare batches, tanks, cages, or different years of stocking, and those reports may lead you to intelligent conclusions (which feed to use, when to stock, which supplier is best, which location is most suitable).

Usually, we just need to know what kind of report we need, create it easily, and use it regularly (daily, weekly, monthly).

The "next level" is basic analysis.

Smart analysis might be able to reveal to you the highest mortality reasons and seasons, the highest feed consumption, and the most successful harvests on a farm.

It is imperative that we make proper visualizations and build appropriate analyses to see the results, otherwise we will miss out on a great deal of information.

Data mining comes next. Here we go a bit deeper, trying to take all our solid data and put it together in one, very well-organized table. Usually, when we look at this table, we see a lot of information. Sometimes even too much. However, it is imperative to “separate the men from the boys” and figure out what information we are looking for, while keeping in mind that this is our very solid data without assumptions, which shows the real status of the farm (provided the data is collected properly and accurately).

When used correctly, data mining can spot trends, outliers, thriving and poor populations, among other things. My experience has shown me that through data mining, some significant decisions have been made. These decisions, such as determining the most appropriate average weight for harvest, predicting harvest distribution, identifying wrong and inaccurate samplings, screening out cages that did not perform well, and many more.

The next level of data usage will require a higher level of knowledge, as well as the use of analytic tools, such as Power BI, and some modelling algorithms.

Before getting to the last two levels, I need to reveal Two things are most critical to keep in mind when dealing with data:

The first is GIGO, which stands for Garbage In, Garbage Out. We will get wrong analyses, wrong reports, and wrong models if we collect inaccurate data, register incorrect information, and record inaccurate data.

Second, data collection is critical, and keeping it consistent without "Black Holes" simplifies the job. In addition to making the task easier, the right tools will also enable the task to be accomplished. Data loss will result if the tools are not suitable for the mission, since the team won't be able to use them.

And, last but not least, we have two levels of data usage:

Usually, this is reserved for medium-large companies, who have collected a large portion of data. This amount of data is usually difficult to view in simple tables and reports. The very big advantage of using BI tools such as Power BI is the fact that it makes connections and relations between different variables. It also demonstrates the effect of one parameter on the other in a very illustrative way, or in very clear data tables.

If we choose to model the farm, i.e., create customized growing and feeding models using AI, based on the data we collected, we will expect to get the most accurate predictions, the finest and most fitted feeding and growing tables, and the largest benefit in that case relates directly to saving money:

- Accurate feeding tables – saves waste of feed

- Accurate models help us to estimate biomass in the farm and in each cage or tank

- Accurate models help us to predict harvesting, by means of timing, sizes, distribution and all the relevant aspects that will help the marketing department to communicate with the clients

These are tools and information that cannot be overstated, and smart use of them will, no doubt, save the company a great deal of money.

I believe that it is very clear that a simple and easy task that can be done, such as gathering day-to-day data from fish farms, is an essential tool to improve performance and profits, and to take the farm to the next level by converting the data into knowledge and moving from guesswork to data-driven decisions.

If you are a fish farmer, and you find it interesting, but you feel you do not have the right knowledge or tools, please contact me and I guarantee support will come.

Nir Tzohari +972-54-9799284

3 views0 comments


Post: Blog2_Post
bottom of page