Horizon-LEAD / lead-platform-doc

Documentation for the LEAD platform
0 stars 0 forks source link

LEAD Platform Documentation

Visit the platform or create an issue requesting for access.

Introduction

The LEAD Platform has been developed to enable the execution of models assisting the stakeholders in the decision making process and the optimization of operations and cost.

Basic Entities

There are four main entities in the platform.

A model consists of a containerised code accepting inputs and providing outputs. A scenario consists of a sequence of models that are going to be run together with each model providing its output as input to the next model. Finally, a simulation is an execution instance of a scenario and contains the information regarding the actual user-provided inputs for each model of the scenario that finally produce the scenario outputs.

Registration

The registration is exclusively though an invitation system. A user can request access by creating an issue on this repository and providing an email to be used for the registration. Subsequently, the user is going to receive an email with an account activation link that will redirect him to a form where he can complete his registration. The account activation link expires after a week.

Usage

Model Creation

From the Model Library menu the user has access to the models available in the platform. Pressing the plus sign (+) redirects the user to the model creation interface. There the user should provide the necessary metadata for the model execution.

Initially the user should provide metadata such as a name, a version and a description for the model. Additionally, the model can be tagged with one of the existing types or with one that the user will create. It is also suggested that a documentation link or source code repository is provided under the Model URL field. Finally, it is necessary to provide a link to an open container registry where the model container image is uploaded.

By checking the "Shared" checkbox the user sets the model to be available to all the platform users.

If the model is used as a connector between two other models, then the "Use As Connector" checkbox should be marked. In the interface that will appear, the user is promped to select the two models in question.

The user can then create the draft model. At this stage, the metadata of the model are created but we have not yet defined any input or outputs for the model. This functionality become available now.

Two new lists appear where the user must declare the input and output variables of the model. By pressing the plus sign (+) of the input data the user is redirected to the model variable creation page. Some generic metadata must be provided such as a name and a description.

Due to the containerized nature of the models' execution, there are two ways of passing arguments (input variables) to the model (input types). As environment (ENV) variables or as command line input (CLI) arguments. The model owner should be aware of the proper selection of this parameter. In the case of an ENV variable the field Variable Name must match exactly (case-sensitive) the environment variable used by the container.

Scenario Builder

Using the scenario menu, the user is presented with a list of the available scenarios. A new scenario can be created by pressing the plus sign (+). In the scenario builder, the user can specify the scenario name and search for the model(s) that will be included in the scenario from the respective field. The selected model can be then drag-dropped into the canvas of the scenario builder.

To connect two models that have been dropped in the canvas the user must drag from the output dot of a model an connect it to the input dot of the next model.

Then a prompt will display if there are any available model connector for this pair. If so, the user can then choose one and the model connector will be automatically placed in the expected sequence in the canvas.

Simulation Execution

A simulation can be initiated from two menus. Either from the scenarios menu, using the menu of the row of a specific scenario of the table and selecting Run Simulation or from the simulations menu by pressing the plus sign (+). In both cases, the user is prompted to provide a name and optionally a description. Pressing Start Simulation will the redirect the user to the simulation wizard, where the user is requested to provide the necesary inputs for each model in the scenario sequence and initiate the execution.

Data Assets

A Data Asset corresponds to some data that the user need to use in a simulation. Usually it consists of data that are used frequently, have a more static nature, and are large enough so that the user would prefer not to constantly upload them for a simulation execution. A Data Asset can contain multiple data files.

From the Data Assets menu the user is presented with a list of all accessible assets. To create a new asset the user can click the plus sign (+). Then a few generic metadata are provided such as a name and a description of the assets included in this data asset. After the assets creation the user can then use the interface to name, describe and upload the files needed.

FAQ