The global skills and competency framework for the digital world

#1359 Add Machine Learning and Data Science skills change request accepted

Add 2 new skills covering Machine Learning and Data Science and a related change to the existing Analytics skill.

The following suggested new skills come from the SFIA community in Australia and included consultation with experts in the data science field. Special thanks to the NSW Chief Data Scientist Dr Ian Oppermann.

Data science (DATS)

The discovery, selection, development, and preparation of data for training and testing machine learning systems. The application of techniques to assess and address the quality, completeness, relevance and potential bias of training data before it is applied to a machine learning system. The application of training and testing data to machine learning systems in order to evaluate the robustness of the training data.

Level 7 - not applicable

Level 6

Specifies the nature and characteristics of data required for training and testing machine learning system. Ensures that appropriate standards and techniques are applied to the discovery, selection, development, and preparation of data for training and testing machine learning systems. Ensures a consistent approach to assessing and addressing the quality, completeness, relevance and potential bias of training data before it is applied to a machine learning system.

Level 5

Identifies the data required to train and test given machine learning systems. Implements quality assessment and remediation activities on the data.
Applies data to learning machine systems and assesses the outcome. 
Provides insights and recommendations to refine data and improve outcomes.

Level 4

Acquires, prepares and quality checks training and testing data for machine learning systems. Applies data to machine learning systems and analyses the outcomes. Carries out complex data quality checking and remediation, recommending alternative data sources where appropriate.

Level 3

Acquires and prepares data for training machine learning systems in accordance with provided specifications. Carries out routine data quality checks and remediation.

Levels 2 & 1 - not applicable

Machine learning (MLNG)

The development and application of systems that learn from data. Evaluation of trained models for their performance, robustness, and bias, including the selection and use of metrics to examine the outcomes. Diagnosis and remediation of issues, both pre- and post-deployment. Anticipation of the organisational implications of machine learning models regarding ethics, bias, privacy, and data protection. Establishing traceability for the outcomes produces by machine learning systems.

Level 7 - not applicable

Level 6

Oversees the technical direction of multiple teams, including research teams developing new approaches to the design, training, and evaluation of machine learning systems. Sets standards and approaches for the application and traceability of machine learning systems to business problems, and oversees their implementation. Designs and oversees organisational policies on the creation, training and use of machine learning systems.

Level 5

Designs, implements, tests and improves machine learning architectures and systems. Selects techniques based on a breadth of knowledge of the strengths, weaknesses and expected performance of different approaches. Establishes good practice in the development, evaluation, monitoring and deployment of machine learning systems.

Level 4

Given a well-described problem and dataset, can anticipate whether machine learning is likely to provide an effective solution. Implements algorithms developed by others. Advises on the effectiveness of specific techniques, based on project findings and on knowledge of wider research. Contributes to the development, evaluation, monitoring and deployment of machine learning systems. Understands and applies rules and guidelines specific to the industry, and anticipates risks and other implications of modelling, including ethics, bias, privacy, and data protection.

Level 3

Applies existing machine learning techniques to new problems and datasets. Evaluates the outcomes and performance of machine learning systems. Identifies issues and recommends improvements to machine learning systems and the data they are based on.

Level 2

Applies given machine learning techniques to data, under the guidance of technical leadership. Analyses and reports findings and remediates simple issues, using algorithms implemented in standard software frameworks and tools.

Level 1 - not applicable

In conjunction with these, we propose the removal of references to 'machine-learning techniques' in Analytics (INAN).

Current status of this request: accepted

What we decided

Included in SFIA 8 AI and machine learning theme.