Outlier soil samples

What is an outlier?

In statistical terms, an outlier is a data point that goes far outside the average value of a group of statistics while in a more general context, it is an individual that is markedly different from the norm in some respect. From the AgroCares perspective, a soil sample is an outlier when it is very different from the other samples that are captured in our database. The soil scans from the sensor in use are sent to AgroCares database whereby they are compared to many other soil scans within the database. The comparable soil scans obtained are then used to determine the fertility status of the soil being analyzed. In some instances, the soil sample in analysis is not comparable to any of the samples within the database, hence termed as an outlier. In that case, AgroCares database cannot be used to generate reliable results for the particular sample. 

Possible reasons for a sample being an outlier

1. High mineral content in the soil: peat soils, ashes in the soil, volcanic soil (soil type dependent),
2. Depth of soil sample is higher than 30cm,
3. Very sandy soils (beach sand),
4. Soil collected in a forest area (high content of organic matter)
5. Sample preparation, main issue would be if the soil was slightly moist, not properly dried.

What to do when a sample is an outlier after the first set of scans. 

Each soil scan twill be little bit different. This difference might just be enough to change the outcome from outlier to a normal sample. We determine reliability for each nutrient separately. The more nutrients are within range, the easier it will be to calculate reliable results from the scans. It is therefore advisable to analyze the sample a second time; reprocess the sample (dry, split and mill) and scan again. After the second set of scans are uploaded, the results of all the scans are combined and used to improve on the comparison for better prediction. If it is still not possible to create a reliable result after this, then the sample is a true outlier. 

The sample is a true outlier, what should we tell the client? 

If a sample is an outlier, we cannot give an advice to the client. However, this does not mean that the farmer has a bad soil, it simply means that the AgroCares soil testing technology cannot offer reliable results. Therefore, when this happens we would like you as a service provider to be honest about it to the client. Explain to him or her that the soil does not match with AgroCares database and that we are not able to calculate a reliable advice for their soil. AgroCares believes that reliability is very important and would not want to return results when we are not certain about them. In return, please make arrangements to refund the client incase payments for the service had already been done.

AgroCares approach to outlier samples

AgroCares is in a continuing process of expanding the database. In this regard, the data management system has been designed to keep track of outlier samples globally and upon request, the respective service provider is expected to send the samples to the Golden Standard Laboratory for further analysis under wet chemistry and later included in the database.