Is predictive analytics a good fit for your context?

Before embarking on a PA project, it is valuable to assess whether it is the right approach at the right time for the setting. Here are a some key considerations to determine if PA is a good fit.

Will PA provide new and actionable information?

  • Do we anticipate that the PA findings will provide information that is more reliable, more frequent or easier to understand than current information available?

  • How will findings from PA be communicated and used, and by whom?

  • How frequently do PA findings need to be updated?

Are there plans and buy-in for acting on insights that come out of PA?

  • Is there sufficient leadership and commitment from the institution?

  • Is there capacity for interpreting, communicating and and acting on PA results? How will findings be communicated?

  • Are there plans for how program services will be changed or how new services will be introduced based on PA results?

  • To what extent does decision making and service provision currently rely on data-driven findings? Are data-driven findings in general, and PA findings more specifically, trusted and used? If not, are there plans for addressing practice and culture to support change?

  • Are there structural or process factors that impede or foster the use predictive analytics (such as dedicated data teams) or the use of results (such as dashboard systems)?

Data quality and systems

  • Are there sufficient data that are readily accessible?

  • Are the data sufficiently documented and understood so that reliable measures (variables to be used as outcomes or predictors) can be created for PA?

  • Is there sufficient capacity for investigating and deploying PA with existing analytic systems, software and staffing?

Potential risks

  • Has their been sufficient investigation into identifying potential risks?

  • Are there systems or processes in place to make sure ethical issues are attended to throughout a PA project?

Back to top