What we do;
Before the launching, we test fresh milk samples without any adultarents and get necessary data to observe the required range for each parameter. Then parameter values are tested after adding water content and graph it with
additional water percentage as the independent variable and parameter values as dependent variables.
How it works;
According to the parameter values of a sample, a graphical visualization of each parameters can be obtained to get a better understanding about tested milk. Further, a grade will be displayed according to the quality as A/B/C/D.
The grading system gives not only a grade for milk but also it is very useful to manage milk storage. It is recommended to store only milk with same grade in the same container, in order to reduce the possible wastages.
This system can be improved so as to get an approximate value for fat percentage of milk which is called SNF value. Since the buyers are not interested in that type of parameters, the proposed grading system is sufficient for
the targeted problem.
Background Mechanism of Grading
SNF - Solid Non Fat , CLR - Corrected Lactometer Reading
The Solids-Not-Fat (SNF) means a collection of proteins, lactose, minerals, acids, enzymes and vitamins contents of the milk. It is the total solid content minus the fat content. The total milk solids are the sum of Fat and
SNF. SNF can be calculated using following formula:
SNF = (CLR/4) + (Fat x 0.21) + 0.36
Our system is not highly focusing on SNF percentage or fat percentage. But what this formula shows is Fat and SNF percentages are factors which affects the milk quality. Hence, deviations from fresh milk is considered in our
system to generate the Grade.
Future Improvements with Machine Learning
After the basic circuit is built up, a fresh milk sample is tested for all parameters with respect to additional adulterant percentages. This data set is used as the training data set to build a data model. Then, this model
can be used to predict approximate adulterant percentage and notify to the user using a mobile notification. Since AWS supports jupyter notebook, after integrating this feature, user can get a better understanding about
milk that he is going to purchase.
According to the grade of tested milk, here is a mechanism to change the price for unit volume (1 litre) by the buyer(collector). For each grade he can assign a value in descending order as grade varies from A to D.
History record of how the values varied is also displayed.