Strategic Investment analysis and Scientific Planning/Alerting VRE Specification (Blue Economy VRE#2)¶
- Strategic Investment analysis and Scientific Planning/Alerting VRE Specification (Blue Economy VRE#2)
- Use Cases
- Strategic Investment Analysis
- Scientific / Environmental Planning / Alerting
- VRE Design
- Use Cases
Strategic Investment Analysis¶
Strategic Investment Analysis is a principal use case for T6.2 and its functionality is delivered by the Strategic Investment Analysis VRE. The end purpose of this VRE is to enable users to locate an area where an investment could be optimally placed. The beneficiaries of the provided functionality are Business Executives of the Aquaculture industry, policy makers, students in both Aquaculture and Economic sciences, and investors.
The VRE will provide one or more economic performance analysis models which may include in their definition the computation of investment performance indicators, such as Internal Rate of Return (IRR) and Net Present Value (NPV).
The users will be able to select one such model according to their preference. Subsequently, a user should select a number of parameters which will be used as input to the model. Among such parameters are the species involved in the aquaculture, the type of the aquaculture and the geographic area under consideration. Financial limitations, such as the location of logistics and other constraints, may also apply and might possibly be reflected on the model as additional parameters. The termination conditions of the model are also settled at this point. Lastly, among the most important parameters for a model are the data sets it will run upon.
All the aforementioned parameters are able to be named, saved, recalled and shared with other users by employing suitable tools provided to the VRE by the infrastructure.
After the model and its input parameters are selected, the user is able to request its execution. The model is the run on the background, and the user is notified upon its completion. The results of the computation are linked with the input parameters and can also be recalled using the functionality of the VRE. The results can also be shared with other users in a similar way with the sharing of input parameters and be visualized on a geographical map using the most suitable visualization method for each case, for example single points or connected paths. In addition, results can be combined with other results and visualized on the same map or graph.
The purpose of the use case scenario described above is to produce a specific result which answers the question “Where an investment case is optimized?”. An alternative scenario under the Strategic Investment Analysis VRE is to run the model with its parameters as described up until now, but on the entire area of the map. The end result is a coloring of the map with results obtained from the execution of the model. However, great care should be exercised for the definition of such scenarios, as the risk that computational needs might explode is present.
Other related use cases and models to the one described above are: Risk minimization, the combination of environmental risk with economic parameters, or the combination of new investment with existing investment.
Scientific / Environmental Planning / Alerting¶
The second major use case for T6.2 is Scientific/Environmental Planning and Alerting. The corresponding VRE’s purpose is to enable users to locate an area where a particular phenomenon, as described by the model, gets maximized or minimized. The user is then alerted when the values associated with the phenomenon meet certain criteria, like being in a particular range, increasing, decreasing or otherwise fluctuating according to user defined rules.
The target users of this use case are Earth Observation Scientists, policy makers and students of Earth Sciences. More details are to be included in subsequent versions of this document, following collaboration with the related Work Packages, namely Blue Assessment (WP5) and Blue Environment (WP7). Alternative use cases and models will also be co-developed or selected in collaboration with the scientists involved in Blue Skills (WP8).
Skill Building and Training is the third major use case for T6.2. This use case is supported by the corresponding VRE and involves students using VRE models for training so that they can explore the variation or impact of a change of one or more parameters over a case.
The target users of this scenario are both instructors and students. Instructors define pre-cooked examples, which include models, initial parameters and results. Student then modify one or more parameters, as defined by the example, run the case and get their results. Students will be subsequently able to report on the impact of their modifications on the model and evaluate their findings.
The following is a list of predefined indices that will be investigated as models for the financial evaluation of an investment. At least two of those will be offered to VRE users.
Net Present Value (NPV)¶
The formula for NPV is as follows:
NPV = Sum(1 to t) [(Ct /(1+r)t - C0]
- Ct = Cash flow at period t
- C0 = Initial investment cost
- r = The discount rate (i.e. the interest rate of money from commercial banks), assumed steady throughout the t periods
- t = number time periods to run the investment over
A positive NPV value indicates an investment with potential profit (not assuming the risk factor). A negative, or close to zero NPV indicates a non viable investment from pure economic perspective.
Internal Rate of Return (IRR)¶
IRR is the growth rate that an investment is expected to yield.
IRR calculatuion is based on Net Present value. The IRR is the solution of the NPV formula by setting the NPV to 0 and solving if for the rate r.
Calculation is iterative and might lead to 0 to 2 solutions.
Return on Investment¶
Measure the ability of an investment to generate profit.
RoE = Net Income / Investment
Return on Equity¶
Measure the ability of an investment to generate revenue for each share given to shareholders.
RoE = Net Profit / Shareholder Equity
* Net Profit = revenue - expenses
* Shareholder Equity = Assets – liabilities
Social Return of Investment¶
ROI is calculated on the basis that extra-financial values are transformed into monetary terms. This latter transformation is domain-specific and can be the case of field research.
Net Profit Margin¶
Measures the net profitability of a business.
NPM = (Revenue - expenses) / expenses
Return on Capital Employed¶
Measures the profitability of the capital employed in an investment
ROCE = net profit / capital employed
Earnings before interest, tax, depreciation and Amortisation (EBITDA)¶
A financial performance indicator that eliminates the effects of financing and accounting decisions. It measures the income before adding interest, taxes, depreciation, and amortization effects on it.
The roles of users involved in all three VREs are defined as follows:
The VRE Manager is able to
- Add models to the infrastructure
- Add data sets to the infrastructure
- Prepare prerequisite data sets
- Configure model data requirements
- Define infrastructure use policy
- Manage VRE membership/registration
A registered VRE user role can an aquaculture employee, manager, a scientist or a student, according to the use case supported by the VRE in question.
A registered VRE user is able to perform the following operations:
- Define and recall model parameters
- Modify model data feeds
- Launch the execution of a model
- Acquire, visualize and exploit model results
This special third role involves a single user identity which is managed by the analytics platform of the VRE and has permissions to perform system operations such as:
- Import and “consume” data feeds
- Run in parallel or deny the execution of models. The execution takes place on infrastructure-provided computational platforms depending on the nature of the model.
- Account platform usage
- Store model results
Process of registration
The registration process in all three described VREs follows the standard procedure followed in all other VREs within the BlueBRIDGE portal.
The Geospatial Analytics and Alerting Platform provides all the functionality supporting the three major use cases described above. A high-level design of the platform is shown in the following diagram.
The platform consists of the following components:
High performance back-end
The high-performance back-end (number 1 in the design diagram) It consists of the sum of all components/nodes which enable distributed processing and provided scalability (scaling-out), high availability and fault tolerance to the VRE. Within the high-performance back-end, there exists a Geospatial Node Cluster, Logical Layer management components (numbers 2 and 3 in the diagram) and a tile cache (number 4 in the diagram). Each geospatial node in the cluster consists of a third party Geospatial Server (e.g. GeoServer), a Geospatial Operations Service which supports Non-OGC RESTful operations of the platform, and services for the coordination of the cluster. Service discovery in the context of the D4Science infrastructure is facilitated by the usage of the gCube Information System. Thus, operating in the context of the D4Science e-infrastructure, the Geospatial Node Cluster is considered a resource provided to the platform by the latter. However, the high-performance back-end in its entirety involves T6.2 as well, as the deployment of services developed in the context of the latter is required for the operation of the platform. For the purpose of authorized access during communication of the High performance back-end with infrastructure services, the gCube Authorization Framework is exploited in the context of the D4Science infrastructure.
Physical layers are replicated across cluster nodes. The set of physical layer replicas comprises a uniquely identifiable logical layer, which is managed by the Logical Layer Broker and Logical Layer Manager Services.
The tile cache will also be a separate service within the high-performance back-end, as it will operate on top of all geospatial nodes.
The analytics layer (number 5 in the diagram) is a VRE service in its own right, as it provides all user-supporting business logic. 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.
Since accounting is an important service for the Geospatial Analytics and Alerting platform, the Analytics Layer shall make heavy usage of such services. For this purpose, in the context of the D4Science infrastructure, the gCube Accounting Service is to be exploited.
External Processing Infrastructure
The execution of analytics models takes place on external computational infrastructures (number 9 in the diagram), implemented in the context of the relevant WPs and provided to the platform by the D4Science e-infrastructure.
The platform employs a centralized configuration service (number 6 in the diagram)whose purpose is to store and provide access to all configuration and state related to the management of logical layers, virtual layers and models (as defined employing the analytics layer) and all user preferences.
This service is expected to be general-purpose and, thus, be provided by Blue Commons.
User Interface elements
The user interface of the platform (number 10 in the diagram) 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 Platform Management UI, which targets VRE Manager and the User Project Management UI, which serves end-users.
Datasets examined for the needs of the VRE are the following:
- Datasets coming from Blue Economy VRE#1, related to aquaculture production models. The access policy is defined by the respective VRE. The datasets are expected to be in the form of georeferenced tables. (one area per table)
- Additional data coming from VRE#1 stakeholders, referring to. Those datasets in their original form will remain private to the stakeholder. Aggregate data (e.g. market rates and wages) will be offered to the entire VRE.
- Datasets containing socioeconomic indices coming from Eurostat, giving coarse statistics for indirect (according to preliminary index design) calculation of monetary / labor / transportation costs and availability. Those datasets are expected to be direct geospatial layers (shape files)
- Datasets containing environmental / earthobservation data coming from FAO Geonetwork (http://www.fao.org/geonetwork/), INSPIRE, ECMWF (http://www.ecmwf.int) . Those data will be open access. The formats vary from vector shapes to rasters and tabular georeferenced data.
- Only authenticated access is supported
Geospatial Analytics Layer Node¶
- At least one node, two for minimal redundancy. The node is not HPC oriented as it is a broker to other services
- Apache Spark: 10+ nodes of 8GB RAM 2-Core
- Storage: 50+ GB for all data types
- 2-10 instances
- 2+ instances
- 1 instance, 50GB+ of storage
- Information System
- Configuration Service
User Interface Elements
- Platform Management UI
- User Project Management UI
- VRE management portlets
Geospatial Analytics Service
- At least one instance
Geospatial Operations Service
- One instance per geospatial cluster node
Logical Layer Services
- Logical Layer Broker: At least 1 instance
- Logical Layer Manager: At least 1 instance
- 1 instance
- VRE Managers: 3-5 per VRE
- VRE End-Users: 10+ per VRE
A geospatial analytics model is defined by a particular fitness or cost function and is potentially accompanied by model training data or other base data.
Its input consists of input parameters, some of which might be optional, which correspond to virtual layers, as defined by the platform.
It responds with vectors of values, among which one defines the fitness/cost in numerical form.
The system invokes the function by supplying to it data that correspond to a particular point in geo-space, time or input parameter space. The latter is valid if the search applies to model parameters too.
The models corresponding to the optimization scenario of the Strategic Investment Analysis use case, mainly run following a probabilistic approach. Consequently, algorithms such as simulated annealing, random walk and genetic algorithms are employed. The algorithms targeting the charting use case must be global in nature, so they run over the entire search space.