Softer aspects beyond tools, techniques and methodologies determine success in analytics projects. The following are critical.
Simplicity:
- Articulate value, process and results in business speak – R2 and confusion matrices may confuse business stakeholders.
- Improve solution ‘deploy-ability’ – focus at the outset on how to implement the analytic solution and use cases from end user perspective to adapt the solution.
- Create friendly and intuitive interactions from the end user perspective.
Diversity:
- Develop people with diverse educational and cultural backgrounds in the team – encourage differences of opinion within boundaries of conflict.
- Develop algorithmic diversity and reduce bias / dependencies on a fixed set of solutions.
- Explore diverse data types and data sources to enrich the solution.
Trust:
Generate reproducible results – Anyone using the same inputs and code as you should be able to arrive at the same results. As an example, use set.seed and equivalent for reproducible results where applicable.
Verify solution using multiple sources – Include independent expert opinions where possible to add perspective to results.
Automate data acquisition touch-points – Reduce chances of error or variation with controls in data acquisition.
Decision Radius 