1. Data Ingestion/integration
1.1 Ability to ingest data from multiple data sources (both internal and external) with no changes to source systems.
Functionality can be provided by configuration of current version of solution.
Akumen supports an open architecture for data ingestion and can ingest data from many sources in various formats through Datasource Manager. This feature currently allows time-based data sources to be configured by uploading files (e.g. .xlsx and .csv). A table is created to store the data within the Akumen database and make the data available to models/apps. Datasource and calculation nodes then facilitate selecting and utilising data throughout the models/apps.
The next iteration of Akumen’s calculation engine will support non-timeseries based data sources. It will allow feeding of data into arrays to allow array-based calculations. This gives users the ability to perform complex tasks with far greater ease.
No changes are required to source systems or their outputs, data can be reformatted within Akumen if/when necessary.
1.2 Ability to use different connection types (API/SDI etc) and frequency of data refresh (real-time/scheduled etc)
Functionality can be provided by configuration of current version of solution.
Sources configured in Datasource Manager can also be set up to be updated via API. Akumen offers two REST API options for connecting to external tools and platforms: Swagger and GraphQL. Akumen also supports the development of custom interfaces as part of the core features.
Akumen’s APIs allow data to be refreshed by pushing data into Akumen. Near real-time data ingestion from data historians is available, with some dependency on the amount of data being transmitted.
Akumen is currently only built to run models with a minimum of hourly granularity. However, more granular data, e.g., minutes, can be ingested into Akumen, aggregated into the required reporting period, and then run through the model.
1.3 Ability to assess data completeness and advise of issues
Functionality available out of the box in current released version of solution.
There are different options for assessing and flagging data completeness and quality. The system implemented depends highly on the requirements of the individual model/application. Options include:
Akumen supports the concept of data reader and data cleaning as part of the standard data ingestion pipeline.
Datasource Manager itself also provides multiple options for error handling during uploading and ingestion.
Python can be used to perform complex data cleaning and review if required, this can be done prior to ingestion, after ingestion and even after the execution of a DDM.
DDM calculation nodes can be used to detect missing data and flag data outside acceptable values.
1.4 Ability to add manual data source or data overwrite if there is an issue
Functionality available out of the box in current released version of solution.
Akumen fully supports the ability to manage different data sets and the data pipeline can be configured to support both manual and automatic depending upon the needs of the customer.
Manual data sources can currently be added and overwritten in two key ways:
Datasource Manager allows Excel files to be uploaded, including .xslx and .csv. Once the data source is initially set up, data can be overwritten or appended to, as necessary. Updated data will automatically flow through to update calculations.
Numeric and timeseries nodes allow users to manually enter data within the models/apps. Once the nodes are linked to calculation nodes, the data can be updated/overwritten, and the new value will be used in the calculations.
1.5 Ability to manage source data (add/remove)
Functionality available out of the box in current released version of solution.
The Akumen Datasource Manager allows users to manage data sources within their Akumen tenancy. DataSource's provide the ability to ingest both CSV, Excel, and JSON files after which the data can then be utilised in models/apps.
The next iteration of Akumen’s calculation engine will introduce a feature called Arrays. This feature will allow feeding of data through Datasource Manager into Arrays to allow array-based calculations. This gives users the ability to perform complex tasks with far greater ease.
Where data has been entered directly into a model/app through Numeric, or Timeseries nodes, these nodes can easily be updated, or deleted and replaced with new nodes where necessary. This can be done both manually through the Akumen interface, or via API (e.g., using a source spreadsheet). Best practice model design requires that nodes containing manual data are clustered together on specific Input pages within the tree-structure. This makes it easier to review and manage large amounts of manual source data.
Nodes can also be changed between type which removes the need to reconfigure subsequent calculations. E.g. A Numeric node representing an average daily production figure is swapped to a Timeseries node with planned daily production. Calculations referencing that node automatically find the correct product for the current day without the user needing to change the calculations.
Complient with the current version of the software.
1.6 Ability to push calculation outputs into other systems for use (i.e. SAP for payment processing)
Functionality can be provided by configuration of current version of solution.
Akumen offers two options for pulling data into external tools and platforms: Swagger and GraphQL. By integrating Akumen with Power BI, organizations can unlock valuable insights and drive data-informed decision-making.
Models can also be manually exported to Excel. Formats can be preconfigured so that exports are automatically formatted in a way that meets business/user expectations. E.g., Monthly Production Report format created to match existing format that users are familiar with.
2. Fact Tables/Business Rules
2.1 Ability to store and view/audit business rules and contractual obligations for the calculation
Functionality available out of the box in current released version of solution.
Business rules and contractual obligations can be recorded throughout the model as appropriate.
These are typically captured within the design and flow of the models themselves. E.g., Mine X’s train loadout operates at a maximum of 10ktph for 16 hours each day, these rules are each represented by a Numeric node. Any relevant calculations, e.g., daily dispatched tonnes, will visually reference and be limited by the Numeric nodes.
All calculations and logic information are freely available for other users with model access to see and investigate. The visual nature of Akumen Dynamic Driver Models helps non-technical users to understand and audit models. Auditing is further assisted by features such as downstream & upstream links and nodes being automatically highlighted when a user hovers over a node. Key changes to nodes are also automatically recorded in the Change History feature.
Akumen also has Label nodes (text boxes) that can be added through the model to explain rules or obligations as needed.
2.2 Ability to change and approve changes to business rules and contractual obligations for the calculation (with an auditable change log)
Functionality can be provided by configuration of current version of solution.
The approval feature of Akumen allows for nodes to be approved by users. Approval is then visually indicated on the node and remains until the node value/calculation is changed. Approval does not prevent the changing of values/calculations, nor are unapproved nodes restricted from use, it is a soft control only.
3. Calculations
3.1 Define, manage and store multiple calculations (standard and adhoc)
Functionality available out of the box in current released version of solution.
Akumen fully supports the ability to define, manage and store multiple calculations. A unique feature of Akumen is the use of natural language expressions to improve the user’s ability to read and understand calculations.
Akumen is also set up to support complex calculations that are then built on sheets that we can then look to templatise as standard calculation/work sheets. This unique capability allows the use of standard templates that can then be used access the business or across multiple sites.
Akumen’s Calculation node takes inputs from different nodes and use them in expressions to produce an output value. These expressions are generated using a simple language based on Excel formulas, but unlike Excel, the expressions reference the names of other nodes rather than using cell references.
A future version of the Calculation Engine that drives Akumen will include Python and Component nodes. Python nodes allow for more complex calculations to be run as part of a model. Component nodes will help to speed up and standardise model development by reusing common calculations, rather than having to rebuild them for e.g., multiple assets.
In Akumen, a single model/app can have multiple pages that may be organised in an intuitive tree structure. These pages can be used to store and organise hundreds and thousands of calculations well beyond the capability of Excel based models, simply.
Models themselves can also be used to separate calculations e.g., a series of calculations that need to be used as an input for several other models can be split into a model which then makes the outputs available to other models through the Asset Library. This makes it easier and repeatable.
3.2 Visually understand what data is used in calculation and the origin (data lineage)
Functionality available out of the box in current released version of solution.
The different types of nodes in Akumen are coloured-coded to allow easy identification of their primary function. E.g. A Numeric node is green, a Timeseries node is yellow. Nodes can also have their background colours changed from the default grey to any custom colour. This feature can be used by model builders to visually communicate things such as status (e.g., draft/approved) or business logic (e.g., all costs related to maintenance coloured blue).
Additionally, if the user hovers over a node, the links to all other nodes it is connected to will be highlighted in red (upstream links) or blue (downstream links). This includes indirectly linked nodes. Node properties also include a list of links to source and destination nodes.
Asset Parameter nodes can be used to get values stored in the Asset Library and use them in the Driver Model. These are often common data points that may be used across multiple models and the data they contain can only be updated from the Asset Library (excepting when adjusting within Scenarios). The Asset Library is effectively the ‘one true source' for specific pieces of information.
3.3 Ability to visually design calculations with a no-code, low code approach
Functionality available out of the box in current released version of solution.
Akumen’s interface allows users to visually design models and calculations using a no/low-code approach. Both technical and non-technical users are empowered to create, modify, and implement complex calculations without the need for extensive coding knowledge. By leveraging the intuitive drag-and-drop nodes, users can integrate data sources and generate insights, all while ensuring the platform remains accessible and user-friendly.
This approach not only accelerates the decision-making process but also democratizes data analytics, making it accessible to a broader range of stakeholders within an organization.
3.4 Ability to run calculations when required and re-run if needed (to allow for data quality processes and controls to be actioned
Functionality available out of the box in current released version of solution.
The user can run and re-run calculations as required, as well as change or update data between runs.
3.5 Ability to visualise / package up outputs for approval
Functionality can be provided by configuration of current version of solution.
Akumen supports the ability to approve outputs through its security model.
Akumen offers several visualization and reporting options, both internal and external:
Akumen is set up to work with Power BI out of the box. This is done by an API call, requires no configuration in Akumen and very little configuration in Power BI. If access to models/scenarios for users building reports needs to be limited, accounts with limited access to specific models/scenarios can be created to facilitate the API calls.
Any other tool that can make REST API calls can be used to build reports on data from Akumen.
A new end user scenario management interface is also in development/design and is currently expected for release in 2023. This will include a range of intuitive visualization options and will be targeted towards users who wish to view or compare model outputs or reports, but don’t require access to the model itself.
Note: We have found that many businesses prefer to use their existing approvals process for outputs. Akumen has been designed to make these outputs easily available in a variety of formats so that they may be slotted seamlessly into existing processes. This has also provided significant benefit to clients by reducing change management and increasing user uptake.
3.5 Ability for business stakeholders to approve calculation outputs for use (in future calculations, reporting or processing)
Functionality can be provided by configuration of current version of solution.
As above, we have found that many businesses prefer to use their existing approvals process for outputs. Akumen has been built to provide easy access to outputs to facilitate these external processes.
Akumen’s Output Protection feature allows models to be locked so that inputs or calculations cannot be changed.
Executed models and their outputs can be access controlled to only allow certain people or roles to make changes.
3.6 Ability to securely store approved calculation outputs for use (in future calculations, reporting or processing
Functionality available out of the box in current released version of solution.
Outputs from the latest run of each model/scenario are stored in the database and are available for use by other models. If Output Protection is enabled, the model will be locked from any changes. Users can utilize the approvals feature to visually record calculations that have been approved.
3.7 Ability to control access to different calculations based on permissions
Functionality can be provided by configuration of current version of solution.
Akumen currently only allows model-level and scenario-level permissions. If users need to have access to only certain groups of calculations, this can be achieved by splitting the calculations into different models. The models can be linked as required via the Asset Library.
3.8 Ability to lock approved calculation outputs, so no further changes can occur
Functionality available out of the box in current released version of solution.
Individual calculation outputs cannot be locked (i.e., prevented from further changes), but Akumen’s Output Protection Feature preserves each model run result when enabled.
An example of this is a business with a model representing their long-term forecast. A user runs the model for a given month and then enables Output Protection to lock the model indefinitely. If the outputs fail the business’s approvals process, the user can disable Output Protection, adjust the model as necessary, rerun the model, then re-enable Output Protection to lock the new outputs.
When preparing to run the model for the next month, the user creates a clone of the original model thereby preserving the previous month’s model in the approved state.
Executed models and their outputs can be access controlled to only allow certain people or roles to make changes.
4. Metric Visualisation and Reporting
4.1 Ability for business stakeholders (with the right access) to view metric outputs, visualise trends, comparisons (i.e. actual to plan) and consolidated views with other metrics
Functionality can be provided by configuration of current version of solution.
Models have inbuilt trend/charts to allow business stakeholders to view comparisons between different metrics or the same metric across scenarios. The ability to view and edit the model is controlled by permissions on the role the user is a member of. Permissions can be locked down at the model level (cannot see the structure of the model) or the study level (can see the structure of the model, but only for certain studies)
4.2 Ability to product reports for different stakeholder groups (internal and external)
Functionality can be provided by configuration of current version of solution.
Akumen has a rich API framework, allowing access (with the correct permissions) from both internally to Akumen (models can be “daisy chained” to feed data between them via API) and externally to Akumen, such as PowerBI or additional modelling tools. This capability allows different stakeholder groups to use their reporting tool of choice to build rich reports that could contain results data from multiple Akumen models together in the same report, right up to building complex dashboards in PowerBI covering Akumen and other systems.
A new end user scenario management interface is also in development/design and is currently expected for release in 2024. This will include a range of intuitive visualization options and will be targeted towards users who wish to view or compare model outputs or reports, but do not require access to the model itself.