The samples are beach litter inventories following the relevant protocols listed here:
These samples represent the amount and type of trash present on a given length of shoreline at a precise location along a designated Swiss river or lake.
The fixed samples are taken at locations that are no greater than one hour by bycycle or walking from the closest public transport stop.
There is a set of twenty locations that have fixed interval samples, this represents the survey results of different lakes on the river.
One off samples are used prmarily but not exclusiveley on connecting water ways.
Different groups and people will have different results, this is a common attribute. We want to be able to differentiate between surveyors.
The results of each sample is a collection of counts. Each count represents a set of like objects with distinguishing attributes (date, location, code) but at least one common attribute. For example:
y = sample_result = 2G79 = "The object G79 was counted twice"
{yi, ... yn} = a set of G79 results on different days or different locations or a collection of different objects at the same location.
y is always greater than zero that is $ 0 < y \leq max $
Length and area information are given with each sample_result. That way object quantities can be expressed as a function of length or area. A sample_result is always positive so you can just divide the sample_result by the attribute:
ypcs = sample_results = 2G79/L; L is the attribute length.
Which means that ypcs is a real number greater than zero.
Where a and b are min/max elements of the survey_result, m is some point that cuts the set in half and f(y) is a density function.
\begin{equation} Pr[a < y < b] = \int_a^b f(y)dy \end{equation}Survey results can be grouped according to the categorical attributes of each result. For example survey results can be grouped according to name. Name refers to the official name of the lake or river the location is situated on.
This will help idenfity density differences of similar objects at different locations, giving an idea of flow and transport of objects.
The changes in the PDF of key objects at different locations will be compared.
The project cycle is roughly two years. The first recorded sample of this type on Lac Léman was Novemeber 15, 2015. Over 80 samples were taken on Lac Léman between Nov 2015 and Nov 2016. A second project was intiated and another cluster of samples was taken on Lac Léman (As well as 14 other lakes) between Apr 2017 and Apr 2018.
The current project will create a third cluster of points in a twelve month period, giving us three defined sampling periods in six years.
The changes in the PDF of key objects and locations will be compared between the three sampling periods.
The sample results are count values. Therefore the regressors have either a postive or negative effect on the count value for an object or a location.
The independent variables used for regression analysis in this project shall come from publicly supported domains of the type, .edu, .admin.ch etc...
This is a small constraint in favor of continuity and reliability.
Certain regressors may not apply to all sample results.
Independent variables of interest: