Interpretation
Use filters and review checkpoints to manage the reduction and elimination of bias (e.g. linear bias, confirmative bias, confidence bias).
Improvement Planning
Practices-Outcomes-Metrics (POM)
Representative POMs are described for Interpretation at each level of maturity.
- 2Basic
- Practice
- Make decisions that are primarily based on results that have documented assumptions, that use the correct data, and the appropriate statistical model.
- Outcome
- There is a basic level of confidence that decisions are based on objective criteria.
- Metric
- Number of decisions backed by data quality, the correct statistical model, and documented assumptions.
- 3Intermediate
- Practices
- Develop a standard decision matrix that includes reliable data capture, documented assumptions, and statistical models.
- Include mechanisms that allow for decisions to be changed when new data becomes available, and that reduce bias from confirmation, over-confidence, and overfitting traps.
- Outcome
- Decisions that are based on the results of data analytics are improved, can be easily updated, and bias is reduced.
- Metrics
- Percentage of decisions where bias is reduced.
- Percentage of decisions updated as new data becomes available.
- 4Advanced
- Practice
- Develop an advanced decision matrix to ensure that: — the data is captured for a specific purpose. — uses an appropriate statistical model.- has documented business assumptions.- can be quickly modified to take account of new data. — has eliminated bias, as far as is practical/cost-effective.
- Outcome
- There is a very high level of confidence in decisions based on the output of data analytics, and that they accurately reflect the most up-to-date information and business requirements.
- Metrics
- Percentage of decisions that accurately reflect future requirements.
- Percentage of investment decisions with a proven return on investment.
- 5Optimized
- Practice
- Use feedback and the latest research available that includes vendors and industry best practices, to drive continuous improvement in how to reduce bias and improve decision-making.
- Outcome
- Research and feedback continually improve both the interpretation of results and decision-making.
- Metric
- Number of implemented changes that have led to improved decision-making.