Performance evaluation, bench-marking and decision making in aquaculture VRE Specification (Blue Economy VRE#1)

Use Cases

In this section, we describe the use cases of VRE#1 “Performance Evaluation, benchmarking and decision making in aquaculture”.
Specifically, for each group of VRE’s potential users a use case is analysed in terms of its requirements, the appropriate steps to implement it and the user roles. Thus, a use case for Fish Farm Managers to make accurate production plans is depicted. As well as, a use case for Business Executives of Aquacultures to design the financial forecasts of their enterprise is presented. Finally, use cases for researchers, scientists, trainers and trainees are also illustrated.

Despite the alternative perspectives and objectives that VRE's user can have on the usage of the service, there are the following discrete steps in all of the use cases:

  1. Define the scenario
  2. Configure the parameters of the scenario
  3. Gather and upload the appropriate dataset(s)
  4. Create a performance analysis model
  5. Save the model and dataset(s)
  6. Perform evaluation of growth by creating what-if scenarios
  7. Save what-if scenarios

Use Case: Production Planning

To create accurate and feasible production plans is a significant process for every aquaculture and depends on numerous physical, biological and environmental factors. In this case, a Fish Farm Manager needs to determine the best month to stock a particular number of fishes of specific specie. Particularly, a user defines the scenario, determines the input parameters, uploads the appropriate datasets and creates a model to evaluate the crucial Key Performance Indicators for the growth. Then, the user utilises the produced model to create what-if scenarios and chooses the best of them. It is worth to note that researchers and scientist of aquaculture sector can utilise this use case for research/scientific purposes, so as to examine the behaviour of Key Performance Indicators for the growth under various circumstances.

For simplicity reasons, let suppose that a Fish Farm Manager wants to check whether is better to stock 150,000 sea bass on June or September. Further details such as the location of the farm and broodstock quality are also known and predetermined. Let suppose that the location is North Rhodes in Greece and the broodstock comes from Johnson Hatcher S.A. without any genetic improvement.

The Fish Farm Manager has to gather data regarding the site oxygen and currents quality and monthly sampling data for Johnson sea bass. For all these data, the user needs to contact the IT department of his/her company, so as to provide excel files containing them.

In order to create a model of performance analysis, the user needs to sign-in as a registered user in the VRE and then the region profile should be defined. Specifically, user defines the profile of the site giving the name of the region, the temperature degrees per half of each month, the oxygen and the current rating of the particular site.

In the next step, user inserts information, such as the name and few comments for the model, the fish species and the broodstock and food quality. Also, user uploads the monthly sampling data file, which is stored in the workspace. After that, a new model is created which is an estimation of the KPIs for the growth based on the historic samplings data. Model is run in background and user is notified notifie upon model run completion. Results are linked to the parameters and can be recalled.

Following, user is able to select and use the performance analysis model to create what-if scenarios. Furthermore, user can compare these what-if scenarios choosing the most beneficial. To do this, the user gives a name for the new what-if scenario and defines the analysis parameters such as the name of the specific model, the amount of fishes (e.g. 150,000), the Average Starting Weight (e.g. 1.35 gr), the date that stock the fishes (starting date), suppose 1st June 2017 and the date that the fishes are going to harvest, suppose 30th November 2018 (end date). What-if analysis is run in background and the results are displayed both in table format and graphically. User can save and recall the results as well as the plots of the what-if analysis.

Finally, the user can either edit the current what-if scenario so as to produce new analysis for the new dates or create a new what-if scenario. In the first case, the new starting and ending date, suppose 1st September 2017 and 31th January 2019 correspondingly, are fed in the scenario and the results, which are produced, could be compared with the previous ones.

The participants in this use case are the VRE users (fish farm managers) and VRE managers/System. The first ones are able to:

  • Define the scenarios;
  • Gather and upload datasets and other model parameters;
  • Create and launch the models for performance analysis;
  • Modify model parameters;
  • Create and launch what-if scenarios;
  • Acquire / Visualize / Exploit results.

The role of VRE managers/System is to:

  • Define infrastructure use policy (VRE manager);
  • Configure model data requirements (VRE manager);
  • “Consume” data feeds for model creation (System);
  • Deny / Run model (System/Execution Platform);
  • Store model (System);
  • Configure what-if analysis data requirements (VRE manager);
  • Deny / Run what-if analysis (System/Execution Platform);
  • Store what-if scenario and analysis results (System).

Use Case: Financial Forecast

The goal of this use case is to illustrate the actions that an aquaculture C.E.O. manager should carry out so as to utilise this service and produce financial forecasts. The user should follow the aforementioned steps and more specifically he/she defines the scenarios, gather the datasets and other input parameters, creates the performance analysis model and eventually performs the evaluation of what-if scenarios regarding the economical indicators. The application does not estimate monetary indicators such as the food and fry cost. However, it provides growth indicators directly related to the cost, like food consumption and biomass produced. It is worth to note that researchers and scientist of aquaculture sector can utilise this use case for research/scientific purposes, so as to contemplate economical indicators under various circumstances.

User should follow the steps of the previous use case (Production Planning), so as to evaluate performance of the growth and produce what-if scenarios for production planning. Alternatively, he/she can co-operate with fish farm manager to request production planning scenarios. Then, the executive manager takes under consideration the results (KPIs estimations) for each one of what-if scenarios and combines them with economical parameters to forecast financial indicators for each scenario. The user could base financial analysis on the following indicators:

  • Biomass produced
  • Average Weight at the harvest date
  • Food consumption
  • Biomass lost due to mortalities

The participants in this use case are the VRE users (aquaculture C.E.O. managers) and VRE managers/System. The first ones are able to:

  • Define the scenarios
  • Gather and upload datasets and other model parameters
  • Create and launch the models for performance analysis
  • Modify model parameters
  • Create and launch what-if scenarios
  • Acquire / Visualize / Exploit results
  • Gather economic data
  • Estimate financial indicators based on KPIs and economic data

The role of VRE managers/System is to:

  • Define infrastructure use policy (VRE manager)
  • Configure model data requirements (VRE manager)
  • “Consume” data feeds for model creation (System)
  • Deny / Run model (System/Execution Platform)
  • Store model (System)
  • Configure what-if analysis data requirements (VRE manager)
  • Deny / Run what-if analysis (System/Execution Platform)
  • Store what-if scenario and analysis results (System)
  • Deny / Run financial forecasts (System/Execution Platform)
  • Store results of financial forecasts (System)

Use Case: Skill Building

Use VRE#1 models and what-if analysis for training purposes, where students can explore the variation of impact of the change one or more parameters over a case. The trainers define and provide some pre-cooked sample datasets. Trainees can use them either to plan the production or to forecast financial indicators. In both options, students can set their desired values so as to create their models and perform KPIs evaluations which are based on these models. Also, they can run what-if scenarios, study the results and report the impacts of different set of parameters.

The participants in this use case are the VRE trainers, trainees and also VRE managers/System. The role of trainers is to:

  • Define training scenarios for production planning as examples to train the students
  • Define training scenarios for financial forecasts as examples to train the students
  • Upload training/sample datasets to utilized by students

The role of trainees is to:

  • Define scenarios for production planning and financial forecasts
  • Set model parameters
  • Create and launch the model
  • Create and launch the what-if scenarios for production planning
  • Create and launch financial forecasts for each scenarios
  • Study the results of the analysis

The role of VRE managers/System is to:

  • Define infrastructure use policy (VRE manager)
  • Configure model data requirements (VRE manager)
  • “Consume” data feeds for model creation (System)
  • Deny / Run model (System/Execution Platform)
  • Store model (System)
  • Configure what-if analysis data requirements (VRE manager)
  • Deny / Run what-if analysis (System/Execution Platform)
  • Store what-if scenario and analysis results (System)
  • Deny / Run financial forecasts (System/Execution Platform)
  • Store results of financial forecasts (System)

KPI extraction (across use cases)

Key Performance Indicators (KPIs) are quantitative measurements that assist an organization to define and evaluate its progress towards its business goals. KPIs represent a set of measures focusing on those aspects of organizational performance that are the most critical for the current and future success of the organization.
In aquaculture the principal Key Performance Indicators relate with the fish growth and fish mortality throughout the lifecycle of an aquaculture ecosystem. KPIs in aquaculture measure the productivity results in terms of fosh growth rate and fish mortality. Usually, these KPIs can be considered as a composition of other lower-level indicators, whose values would be obtained from the databases of an aquaculture organization. The well-known indicators in aquaculture are economical and biological Feed Conversion Rate (FCR), Growth Rate per Day (GPD), Specific Growth Rate (SGR), Suggested Feeding Rate (SFR) and Mortality Rate (MR).

Feed Conversion Rate(FCR)

The most important KPI in aquaculture is the Feed Conversion Rate (FCR) which is calculated from the number of kilograms of feed used to produce one kilogram of fish (e.g. FCR = 1.5 means that 1.5 kg of feed is needed to produce one kilogram of fish live weight). In other words, FCR measures the yield of the biomass with respect to the feed required to produce it. It is worth to note that feed being the primary input cost of the aquaculture process.

Usually, two additional terms are used: the biological FCR and the economical FCR. Biological FCR is the net amount of feed used to produce one kg of fish, while the economical FCR takes into account all the feed used, meaning that the effects of feed losses and mortalities, for example, are included.

Economical FCR is calculated as follows:

Economical FCR

where

Net Growth

The Biological FCR differs from the Economical FCR as it is a function of the gross growth of the biomass and is derived as follows:

Biological FCR

where

Gross Growth

The Biomass of Mortalities is a measure of the loss throughout the lifetime of the batch due to fish mortality.

Growth Per Day(GPD)

Growth per Day (GPD) is a measure of how the fish are growing a a daily base. It is calculated by the following formula:

GPD

where
T1 (Time 1) presents the initial point in time in the lifecycle of a batch
T2 (Time 2) presents the end time point in the lifecycle of a batch
Δt is the time interval between the two dates T2 and T1, measured in days
AverageWeghtT2 presents the average body weight of the fish at time T2 and
AverageWeghtT1 is the average body weight of the fish at time point T1

Specific Growth Rate (SGR)

The Specific Growth Rate (SGR) is the relative growth rate i.e. the growth rate relative to the size (body weight) of the fish. It is calculated by the following formula:

{{thumbnail(Equation_6.png)}}

where
ln(AverageWeight_T2) is the natural logarithm of the final average boby weight of the fish at time T2,
ln(AverageWeight_T1) is the natural logarithm of the initial average boby weight of the fish at time T1 and
Δt is the time interval in days between the dates T1 and T2.

Suggested Feeding Rate (SFR)

The Suggested Feeding Rate (SFR) is the

SFR

where
Clossing Biomass is the biomass at the date T2
Harvest Biomass is the biomass that is harvested in the period between the dates T1 and T2
Opening Biomass is the biomass at the date T1

Mortality Rate (MR)

Fish mortality is a parameter used in fisheries population dynamics to account for the loss of fish in a fish stock through death. Mortality Rate is the ratio of fish deaths in an area to the fish population of that area.

{{thumbnail(Equation_8.png)}}

where LTD mortalities presents the number of fish which are died in the period between the dates T1 and T2.

Performance comparison

One of the services that VRE#1 Performance evaluation, benchmarking and decision making in aquaculture provides to aquaculture end-users is a collaborative working environment allowing them to contribute and exploit anonymized data, for the evaluation of their performance. In this environment enterprises have the ability to benchmark their KPIs against indicators of other companies under similar environmental conditions in terms the temperature and currents in a specific region. A challenge concerns the confidentiality of the data. In more details, the crucial issue is how to handle the datasets, which are provided by aquacultures, without disclosing personally identifying information and losing their anonymity. The confidential aquacultures data will be submitted to anonymization process that means identifiable information will be removed, so that aquacultures remain anonymous.

After the aforementioned process the performance comparison can be occurred. The goal of benchmarking is to assist the aquaculture managers to estimate their companies' KPIs performance against other aquaculture enterprises which operate under similar circumstances, such as region, temperature and currents. Having data from different aquacultures that operate in a specific region, the service can use them to model a "global" general behaviour from the KPIs. Thus, a benchmarking between the KPIs performance of the particular aquaculture versus the "global" KPIs performance could be take place. This performance comparison supplies the managers with the ability to realise the potential margins of improvements that can have so as to make correct and valid decisions regarding their production. In the following use case, the above context of comparison is explained in detail.

Suppose that an aquaculture manager or fish farm manager wants to perform a comparison of his/her company’s FCR against the mean (“global”) FCR of other aquacultures in the same region and under similar environmental conditions (currents, temperatures). The manager attempts to determine if the FCR is worse than the average FCR of the specific region. Having this knowledge, the administrator can examine whether it can produce the same fish biomass spending less amount of food and consequently reducing costs. The steps which a manager has to execute so as to perform benchmarking are similar with those in the aforementioned production planning use case. First of all manager has to define the scenario, gather and upload datasets and other model parameters. Then, he/she has to create and launch the models for performance analysis and modify model parameters. After that, he/she has to create and launch what-if scenarios and then he/she acquires and visualizes the results. To perform benchmarking an additional step is required. The company’s FCR KPI compares with the average FCR, if and only if more than one company operates in the same region. Finally, user can acquire and visualize the benchmarking results.

Users

The stakeholders of the VRE#1 services are aquaculture SMEs, scientists and researchers from institutes and universities relative with the sector and also government organizations and decision makers.
As the objectives of VRE are the providing tools and services to benefit aquaculture enterprises, the majority of the users will come from this sector. The role of aquaculture SMEs is dual as they contribute on one side for the evaluation of their performance and consequently validation the VRE’s services and on other side as end-users of VRE’s services. In particular three SMEs have already chosen as internal evaluators (subcontractors) of the services. They will participate in requirements analysis, in gathering the appropriate datasets and in VRE evaluation and exploitation. A briefly presentation of the three subcontractors exists in the following subsection. Also, we intend to invite a wide audience of aquaculture SMEs to evaluate the services when they will be ready to deploy.

Furthermore, researchers and scientists from Research/Scientific Centers and Universities of relative domains (aquacultures, fisheries, ichthyology etc) will have access to datasets concerning the practices and performance of aquaculture producers, generating new knowledge and evaluating the practical indicators of aquafarming performance.

Finally, government organizations and decision makers use the VRE services for designing investment policies, study environmental impacts and also evaluate the current situation and define policies.

VRE Design

The objective of the VRE#1 “*Performance evaluation, benchmarking and decision making in aquaculture*” is to provide tools and services to end-users enabling them to convert their operation data into knowledge in a user-friendly environment. The VRE#1 system provides the following functionalities:

  • User-friendly environment to manage the Site profile.
  • User-friendly environment to create, modify and store models.
  • User-friendly environment to upload and store datasets.
  • User-friendly environment to create, modify and store what-if scenarios so as to evaluate the performance of crucial KPIs.
  • User-friendly environment to display, recall and store the results of the analysis.

A high-level design of the VRE#1 system is illustrated in the Figure 1. It consists of the following components:

  • User Inteface (UI) & Visualisation Layer: The user interface of the platform will be realized within the infrastructure as portlets conforming to the standards of the BlueBRIDGE project. The UI functionality, at a high level, comprises the Site Management UI, the Model Management UI and the What-If Analysis Management UI. In subsection "User Interface Design", detailed descriptions of these portlets are exhibited.
  • Analytics Layer: The analytics layer is a VRE service in its own right, as it provides all user-supporting business logic. In this level the user inputs and the datasets should be utilized so as to the required models, for the what-if analysis, will be deployed. Service discovery in the context of the D4Science e-infrastructure is facilitated by the usage of the gCube Information System. For the purpose of authorized access during communication of the Analytics Layer with infrastructure services, the gCube Authorization Framework is exploited in the context of the D4Science infrastructure. Moreover, accounting is a desired service for the system, so for this purpose, in the context of the D4Science infrastructure, the Analytics Layer will utilize the gCube Accounting Service.
  • Data Layer: within this back-end layer, there exists a Relational Database in which necessary entities are accumulated. The user input datasets are stored inside the DB schema as Binary Large Objects (BLOB). Also, the created models and what-if scenarios are stored in their own database tables. In subsection "Database Schema", a description of the Database scheme is exhibited.
  • External Processing Infrastructure: The execution of analytics models takes place on external computational infrastructures, implemented in the context of the relevant WPs and provided to the platform by the D4Science e-infrastructure. In order to analyze the performance of the aquaculture activities, a Model Manager calls the suitable R functions, which have been already implemented and stored in the D4Science e-infrastructure. The created model is returned as output to the Model Manager and stored in the Database.
  • Configuration Service: The platform employs a centralized configuration service whose purpose is to store and provide access to all configuration and state related to the analytics and data layer. This service is expected to be general-purpose and, thus, be provided by Blue Commons.

Figure 1. Design of the VRE#1 service

User Interface Design

The application provides three (3) portlets to the registered user (Figure 2):

  1. Site Management
  2. Model Management
  3. What-if Analysis

The first two allow the user to define the context into which the what-if analysis will take place.

Figure 2. The User Interface portlets

Site Management

In Site Management user can create a new site of interest by clicking the Add button. In the following display (Figure 3) user inserts information about the site. Specifically, the oxygen rate, the current rate and the Temperatures per half of each month are required. User has to type the 24 temperatures. For example, Jan A presents the average temperature in the first half of January and Jan B indicates the average temperature of the second half of the particular month and so on. When user clicks on “**Save**” button, then the Site Profile will be stored.

Alternative, when user choose the Region of the Site, then the fields “Oxygen Rating” and “Current Rating” could be automatically completed by receiving the related information from external resources, like those which are described in the section "Resources".

Figure 3. Site Management

Model Management

In this portlet, the users have the capability to manage their model (Figure 4). A list of existing models is displayed that includes the model's name, some comments and the specie of the fish that the specific model is referred to. A user can create a New Model pressing the “**Add**” button or delete an existed model.

Figure 4. Model Management

When the Add button is pressed, the following screens are displayed. The “Model Management” section splits into two parts. In the first part (Figure 5), the user can create a new model giving a name for the model, typing some useful comments and choosing from drop-down lists the name of the specie, the site, the Broodstock Quality and Freed Quality. Also, if the Broodstock is improved genetically, the user has to check the corresponding box (Broodstock Genetic Improvement). In the second part (Figure 6), the user can upload dataset from files (in csv, xlsx format). After that, the user can press the “**Save and Generate Model**” button, so as to save all the input data and create the model. The new model is added to the list of models.

Figure 5. Create a New Model

Figure 6. Upload files

What-If Analysis

In this portlet, the users can make their own what-if scenarios, based on the models that they have already created. A list of existing what-if scenarios is displayed that includes the scenario’s name, some comments and the name of the model that is utilised (Figure 7). When the “**Add**” button is pressed, a new what-if scenario will be created. Also, the user can run, delete and/or edit the existing what-if scenarios.

Figure 7. What-If Analysis

In the following screen the new what-if scenario will be created (Figure 8). The user gives the name and writes some comments for the new what-if scenario. Also, he/she can choose the model in which the new what-if scenario is based on and provides information such as the starting date, the amount of fishes, the starting average weight and the final (target) date. After that, the user can press the “**Save and Generate**” button, so as to save the what-if scenario.

Figure 8. Create a What-If scenario

In the following screens, the user can see the results of the what-if analysis (Figure 9). First of all, the estimation of main KPIs, such as Average Weight, LTD Growth, LTD SGR, LTD Biological and Economical FCR and LTD Mortality are presented. Secondly, graphs are provided to the users, such as Weight Graph, FCR comparing with the global trend of FCR for benchmarking purposes and Food Consumption graph in the specific time period (Figure 10).

Figure 9. Results of What-If analysis

Figure 10. Visualisation the results of What-If analysis

Database Schema

The Relational Database provides the functionality to store, access, retrieve, secure and integrate users data within the database. It contains details about the Site, Region, Oxygen and Current Rating, Species, Broodstock and Feed Quality and stores them in the corresponding entities (tables). The information about the models, which are created based on user’s input, are stored in the entity named “SimulModel”. The output of each model is stored in the entity with the corresponing name (“Fcr”, “Sfr”, “Sgr” and “Mortality”). Furthermore, in the entity “Scenario” the details about a scenario as well as the results of the performance are stored. The Entity-Relationship diagram of the Relational Database is depited in the following figure (Figure 11):

Figure 11. Entity-Relationship Diagram of Database

The database entities (tables) with their attributes as well as their data types are the following:

Region

regionId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
name STRING null

CurrentRating

currentRatingId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
aa INTEGER,
name STRING null

FeedQuality

feedQualityId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
aa INTEGER,
name STRING null

OxygenRating

oxygenRatingId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
aa INTEGER,
name STRING null

Site

siteId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
name STRING null,
periodJanA INT,
periodJanB INT,
periodFebA INT,
periodFebB INT,
periodMarA INT,
periodMarB INT,
periodAprA INT,
periodAprB INT,
periodMayA INT,
periodMayB INT,
periodJunA INT,
periodJunB INT,
periodJulA INT,
periodJulB INT,
periodAugA INT,
periodAugB INT,
periodSepA INT,
periodSepB INT,
periodOctA INT,
periodOctB INT,
periodNovA INT,
periodNovB INT,
periodDecA INT,
periodDecB INT,
oxygenRatingId INT,
currentRatingId INT,
regionId INT

Species

speciesId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
name STRING null

BroodstockQuality

broodstockQualityId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
aa INTEGER,
name STRING null

FCR

fcrId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
simulModelId INT,
temperature INTEGER,
fromWeight DOUBLE,
value DOUBLE

Mortality

mortalityId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
simulModelId INT,
temperature INTEGER,
fromWeight DOUBLE,
value DOUBLE

SFR

sfrId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
simulModelId INT,
temperature INTEGER,
fromWeight DOUBLE,
value DOUBLE

SGR

sgrId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
simulModelId INT,
temperature INTEGER,
fromWeight DOUBLE,
value DOUBLE

SimulModel

simulModelId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
name STRING null,
comments STRING null,
speciesId INT,
siteId INT,
broodstockQualityId INT,
broodstockGeneticImprovement BOOLEAN,
feedQualityId INT,
Dataset1 BLOB,
Dataset2 BLOB,
Dataset3 BLOB,
Dataset4 BLOB

Scenario

scenarioId INT not null primary key,
groupId INT,
companyId INT,
userId INT,
userName STRING null,
createDate DATE null,
modifiedDate DATE null,
name STRING null,
comments STRING null,
simulModelId INT,
startDate DATE null,
fishNo INTEGER,
weight DOUBLE,
targetDate DATE null,
resultsWeight DOUBLE,
resultsGrowth DOUBLE,
resultsEconFCR DOUBLE,
resultsBiolFCR DOUBLE,
resultsSGR DOUBLE,
resultsMortality DOUBLE,
resultsGraphData TEXT null

Resources

Data Sets

External Datasets

Potential external resources to provide VRE#1 services with datasets are the followings:

During the Site selection at the Site Management portal, the system could provide to the end-user the capability to automatically choice the oxygen and current ratings for the specific region. This information will be provided via the connection in the above web sites.

Internal Datasets

The anonymized data, which are owned by the aquacultures, are involved in the cases of performance evaluation (KPIs estimation) and benchmarking analysis. These datasets contain sampling information which are gathered from time to time by Aquaculture companies and hereafter will be called “Sampling To Sampling“ dataset. Sampling is a common procedure in the aquaculture sector, in order to estimate the number and the average weight of fishes in cages/units. This kind of datasets can be generated by monitoring systems used by Aquaculture companies. It is consisted of the following attributes (fields):

  • Unit & Batch
  • Start Date: starting date of sampling
  • End Date: end date of sampling
  • Start Fish No: number of fish at the beginning of sampling
  • Start Av. Wt: fish average weight at the beginning of sampling
  • End Fish No: number of fish at the end of sampling
  • End Av. Wt: fish average weight at the end of sampling
  • Av. Temp.: average temperature during the period between particular samplings
  • Feed Qty: quantity of food given the period between the particular samplings
  • Mortalities: fish mortalities during the particular period
  • Harvest #: number of fish harvest the period from sampling to sampling
  • Harvest Kg: weight of harvest in kilograms the period from sampling to sampling

Using these attributes will be able to calculate useful KPIs for production, such as Feed Conversion Ratio (F.C.R.), Specific Feeding Ratio (S.F.R.), Specific Growth Ratio (S.G.R.) and Mortality Rate (M.R.). The dataset is tabular data of 150 up to 5000 entries usually in Excel format (xlsx, csv).

There isn’t any Embargo Periods and it obeys in Non-commercial Creative Commons (NC) license. These “Sampling To Sampling“ datasets will remain private to the stakeholder in their original form and the access policy is defined by the respective VRE. The specific datasets can be used by any authorised user or researcher of VRE, so as to produce KPI’s tables, such as FCR, SGR, SFR, Mortality tables. Although, for training purposes, stakeholders will provide open-access “Sampling To Sampling“ datasets.

The generated dataset contains the values of the Feed Conversion Ratio (F.C.R.) KPI, which is produced after the statistical modeling that will be done in the VRE at the sampling to sampling dataset. The FCR KPI is one of the majors indicator of growth performance in Aquaculture sector. It is cross-tabular data where the rows present the categories of the Average Weight in bins and columns present the Temperature degrees in Celsius. Usually, it contains 20 up to 50 rows ( (0 gr, 50 gr], (50 gr, 100 gr], (100 gr, 150 gr],… ) and at most 30 columns (Degrees of Temperature, 0o C, 1o C, 2o C,… ). The specific cross-tabular table can be used by any authenticate user or researcher of VRE, so as to produce what-if scenarios about the expected growth. Also, a sensitivity analysis can be made using the FCR table. It is useful at any Aquaculture company and can be used for simulation almost by any Production Management System. There isn’t any Embargo Periods and it obeys in Non-commercial Creative Commons (NC) license and the access policy is defined by the respective VRE.

The dataset contains the values of the Specific Growth Ratio (S.G.R.) KPI, which is produced after the statistical modeling that will be done in the VRE at the sampling to sampling dataset. The SGR KPI is one of the majors indicator of growth performance in Aquaculture sector. It is cross-tabular data where the rows present the Average Weight in bins and columns present the Temperature degrees in Celsius. Usually, it contains 20 up to 50 rows ( (0 gr, 50 gr], (50 gr, 100 gr], (100 gr, 150 gr],… ) and at most 30 columns (Degrees of Temperature, 0o C, 1o C, 2o C,… ). The specific cross-tabular table can be used by any authenticate user or researcher of VRE, so as to produce what-if scenarios about the expected growth. Also, a sensitivity analysis can be made using the SGR table. It is useful at any Aquaculture company and can be used for simulation almost by any Production Management System. There isn’t any Embargo Periods and it obeys in Non-commercial Creative Commons (NC) license and the access policy is defined by the respective VRE.

The dataset contains the values of the Suggested Feeding Ratio (S.F.R.) KPI, which is produced after the statistical modeling that will be done in the VRE at the sampling to sampling dataset. The SFR KPI is one of the majors indicator of growth performance in Aquaculture sector. It is cross-tabular data where the rows present the Average Weight in bins and columns present the Temperature degrees in Celsius. Usually, it contains 20 up to 50 rows ( (0 gr, 50 gr], (50 gr, 100 gr], (100 gr, 150 gr],… ) and at most 30 columns (Degrees of Temperature, 0o C, 1o C, 2o C,… ). The specific cross-tabular table can be used by any authenticate user or researcher of VRE, so as to produce what-if scenarios about the expected growth. Also, a sensitivity analysis can be made using the SFR table. It is useful at any Aquaculture company and can be used for simulation almost by any Production Management System. There isn’t any Embargo Periods and it obeys in Non-commercial Creative Commons (NC) license and the access policy is defined by the respective VRE.

The dataset contains the values of the Mortality Rate (M.R.) KPI, which is produced after the statistical modeling that will be done in the VRE at the sampling to sampling dataset. The Mortality Rate KPI is one of the majors indicator of growth performance in Aquaculture sector. It is cross-tabular data where the rows present the Average Weight in bins and columns present the Temperature degrees in Celsius. Usually, it contains 20 up to 50 rows ( (0 gr, 50 gr], (50 gr, 100 gr], (100 gr, 150 gr],… ) and at most 30 columns (Degrees of Temperature, 0o C, 1o C, 2o C,…). The specific cross-tabular table can be used by any authenticate user or researcher of VRE, so as to produce what-if scenarios about the expected growth. Also, a sensitivity analysis can be made using the MR table. It is useful at any Aquaculture company and can be used for simulation almost by any Production Management System. There isn’t any Embargo Periods and it obeys in Non-commercial Creative Commons (NC) license and the access policy is defined by the respective VRE.

Nodes

A moderate scenario about the usage nodes of the VRE#1 service considers that 100 aquaculture companies are potential users, so 100 User profiles will be needed. The VRE#1 software application has no substantial requirements in CPU cores and RAM. An average estimation about the Database storage is 200 MB per user. As well as, it is needed 30 Rserve pool connections in order to connect Java portlet applications with R project.

The following nodes are expected to be used:
The initial number of virtual hosting nodes needed for the system is :
* 1 for backend services
* 1 for the UI (shared portal)
* 1 for data persistency (shared data base node)

Services

A moderate estimation of the concurrent users of VRE#1 service is at most 30 users. Thus, the following infrastructure services will be needed:

  • 30 Rserve pool connections
  • 30 Databases pool connections

Models

The models and algorithms to perform evaluation and benchmarking of KPIs come from the field of statistical learning theory which is a framework for machine learning. Statistical learning theory has successfully applied in various fields such as computer vision, pattern and speech recognition, bioinformatics, data mining etc. It deals with the problem of finding a predictive function based on historical data. Mostly, the statistical learning models involve learning from a training set of data. Every object in the training is an input-output pair of values, where the input assigns to an output. The learning problem consists of inferring the function that maps between the input and the output, such that the learned function can be used to predict output from future input. Depending of the type of output, supervised learning problems are either problems of regression, if the output takes continuous values, or problems of classification, whether output is a discrete variable. It is worth to note, that prediction within the range of values in the dataset used for model-fitting is known informally as interpolation. Prediction outside this range of the data is known as extrapolation.

In our case, we use regression analysis which is a statistical process for estimating the forms of relationships among independent variables and response (dependent) variable. Also, models of regression analysis can be utilised to understand which among independent variables are related to the dependent variable. Specifically, models take as input features of the sampling to sampling datasets, like Average Temperature, Start Average Weight of fish as well as other information regarding the region and estimate the KPIs tables, such as Feed Conversion Ratio (F.C.R.) KPI Table, Specific Growth Ratio (S.G.R.) KPI Table etc. The well-known algorithms of regression analysis, generalized linear models, locally weighted scatterplot smoothing, generalized additive models, smoothing splines among them are algorithms that we will be utilized to modeling. Furthermore, an algorithm that simulates the growth and estimates the KPIs performance in a specific period of time will be executed. The outcome of this process will be displayed to the end-users via graphs and tables of results.

Figure_2.png (17.5 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_3.png (10.3 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_4.png (21.6 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_5.png (16 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_6.png (10.3 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_7.png (21.4 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_8.png (13.5 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_9.png (24.9 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_10.png (74.3 KB) Gerasimos Antzoulatos, Feb 12, 2016 03:56 PM

Figure_2.png - {{thumbnail(Figure_2.png)}} (11.7 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_3.png - {{thumbnail(Figure_3.png)}} (18.2 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_4.png - {{thumbnail(Figure_4.png)}} (12.1 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_5.png - {{thumbnail(Figure_5.png)}} (7.79 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_6.png - {{thumbnail(Figure_6.png)}} (6.86 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_7.png - {{thumbnail(Figure_7.png)}} (12.6 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_8.png - {{thumbnail(Figure_8.png)}} (7.05 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:26 PM

Figure_9.png - {{thumbnail(Figure_9.png)}} (33.5 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:29 PM

Figure_10.png - {{thumbnail(Figure_10.png)}} (29.4 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:29 PM

Equation_1.png - {{thumbnail(Equation_1.png)}} (5.61 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:43 PM

Equation_2.png - {{thumbnail(Equation_2.png)}} (9.37 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:54 PM

Equation_3.png - {{thumbnail(Equation_3.png)}} (5.12 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:54 PM

Equation_4.png - {{thumbnail(Equation_4.png)}} (7.13 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:54 PM

Equation_5.png - {{thumbnail(Equation_5.png)}} (4.34 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:58 PM

Equation_6.png - {{thumbnail(Equation_6.png)}} (5.38 KB) Gerasimos Antzoulatos, Feb 18, 2016 03:59 PM

Equation_7.png - {{thumbnail(Equation_7.png)}} (7.25 KB) Gerasimos Antzoulatos, Feb 18, 2016 04:01 PM

Equation_8.png - {{thumbnail(Equation_8.png)}} (4.27 KB) Gerasimos Antzoulatos, Feb 18, 2016 04:04 PM

Blue_Economy_VRE_1_Specification.pdf (1.04 MB) Gerasimos Antzoulatos, Feb 18, 2016 05:12 PM

Figure_1_System_Design_VRE1.png (83 KB) Gerasimos Antzoulatos, Mar 15, 2016 01:11 PM

Figure_11_DBschema.png (87 KB) Gerasimos Antzoulatos, Mar 15, 2016 01:11 PM

Figure_11_DBschema.png (87 KB) Gerasimos Antzoulatos, Mar 15, 2016 01:22 PM