Back

Develop and implement machine learning algorithms

URN: TECIS805401
Business Sectors (Suites): IT(Data Science)
Developed by: e-skills
Approved on: 29 Apr 2020

Overview

This standard identifies the competences you need to develop machine learning algorithms. It includes the different approaches to machine learning and their implementation in accordance with approved procedures. Machine learning algorithms are used in a wide variety of applications where it is difficult or infeasible to develop a conventional algorithm to perform the task. This will involve the practical use of software tools for machine learning algorithm development. You will understand how to identify and select tasks suitable for machine learning and formulate machine learning problems in order to address them. Your underpinning knowledge will enable you to develop and test machine learning algorithms. You will be required to select and apply machine learning algorithms to build models for prediction, classification or clustering. You will undertake the process of training and validation in order to develop machine learning solutions. You will be able to assess the performance of a developed model and identify the role of training and test datasets in this process. You will understand the process of training and validating machine learning models. The standard will introduce the concepts of error and bias in model development and their importance in evaluating model performance. This activity can be increasingly found in any sector or organisation. This activity is likely to be undertaken by people working as Machine Learning Specialists or Machine Learning Engineers.

Performance criteria

You must be able to:

  1. prepare datasets from multiple databases and other sources to input into machine learning models
  2. capture, organise and prioritise requirements to describe organisational needs
  3. evaluate datasets to identify quality issues to determine and document an approach to addressing them
  4. translate business and technical requirements into machine learning problems to plan and develop solutions
  5. conduct data cleaning of noisy, incomplete or data with established data quality issues using approved tools and techniques
  6. select and develop data sets, algorithms and modelling techniques required to solve organisational data problems
  7. create analytical models to produce machine learning solutions
  8. evaluate and validate machine learning models to ensure no bias is introduced
  9. apply best-practice techniques for output model testing and tuning to assess accuracy, fit, validity and robustness
  10. design and implement dashboard and automated reporting systems to deliver updates on model performance
  11. develop strategies for model improvement as well as improvements to data and retraining
  12. create and disseminate reports, presentations and other documentation that provides storytelling and description of model development to confirm stakeholder approval for handover to implementation

Knowledge and Understanding

You need to know and understand:

  1. the stages of the machine learning lifecycle and how to apply them
  2. the characteristics of different machine learning methods and models including; supervised learning; unsupervised learning; text mining, reinforcement learning, ensemble learning; predictive modelling; classification models; regression models and clustering models
  3. a wide range of statistical methods and best-practice modelling techniques and how to apply them
  4. the required data cleaning techniques used to improve data quality

  5. the dataset preparation activities that are required in the machine learning process including data collection, formatting, reduction, decomposition and rescaling

  6. how to select and apply machine learning algorithms for classification, regression and clustering using existing libraries

  7. the required machine learning procedures for text data

  8. the steps involved in machine learning output model validation and how to apply them

  9. the variables and features that impact model performance to test and validate output model performance

  10. the factors that impact model validation such as the size of the data set and how it is segmented 

  11. the differences between structured and unstructured data

  12. the required training and testing steps for data sets to produce accurate models

  13. how to evaluate machine learning model performance

  14. the tools, systems and procedures for developing machine learning models

  15. the techniques for identifying and reducing bias in datasets and how to apply them

Scope/range

Scope Performance

Scope Knowledge

Values

Behaviours

Skills

Glossary

Links To Other NOS

External Links

Version Number

1

Indicative Review Date

30 Mar 2023

Validity

Current

Status

Original

Originating Organisation

ODAG Consultants Ltd

Original URN

TECIS805401

Relevant Occupations

Information and Communication Technology Professionals, Software Development

SOC Code

2139

Keywords

Machine learning, algorithms, artificial intelligence