Embedding Application-Specific Analytics
Make use of canned or bundled data analytics by developing trust with equipment suppliers who have embedded domain-specific analytics in their equipment.
Improvement Planning
Practices-Outcomes-Metrics (POM)
Representative POMs are described for Embedding Application-Specific Analytics at each level of maturity.
- 2Basic
- Practice
- Use automation when the problem is relatively straightforward — i.e. for tasks where there is a clear, predefined sequence of steps that can be transitioned from a human to a machine (for example, to filter out negative values).
- Outcome
- The speed and efficiency of data analytics is improved, and staff are freed up for more productive activities.
- Metric
- Number of manual data analytics steps that have been automated.
- 3Intermediate
- Practices
- Use machine learning to determine how certain features of the data are related to the outcomes you might be interested in.
- (Please note machine learning is suitable only for problems that require prediction rather than causal inference, and that are sufficiently self-contained or relatively insulated from outside influences).
- Outcome
- Machine learning eliminates guesswork by gathering and analysing large volumes of data to optimally leverage the resulting insights.
- Metrics
- Number of insights gained by machine learning.
- Value of insights gained from machine learning.
- 4Advanced
- Practice
- Decide where machine learning can be embedded, based on data suitability/quality and the tolerance for exceptions or statistical errors.
- Outcome
- Some embedded analytics are used and are rolled up to organizational data analytics.
- Metric
- Percentage of embedded machine learning used by the organization.
- 5Optimized
- Practice
- Use vendor research and internal feedback to build trust with vendors to both influence development and to provide completely embedded solutions.
- Outcome
- Embedded analytics are used exclusively in areas where trust in the vendor is justified.
- Metric
- Percentage of embedded (vendor provided) analytics.