Avoid Pitfalls of Predictive Analysis
There are plenty of lessons in predictive analysis models, according to a 2011 white paper, “Worst Practices in Predictive Analysis,” by Information Builders, a company that focuses on enterprise business intelligence and Web reporting software solutions. Here is how to avoid them:
• Determine the ROI. When planning to implement predictive analysis, consider the total cost and the anticipated return to ensure the maximum value is achieved.
• Focus on bottom-line initiatives. Create models that provide forward-looking information to help solve specific problems, like a very low graduation rate, or to help achieve certain goals, like improving math performance in at-risk students.
• Prepare the data correctly. Guarantee the most accurate possible results by ensuring all the data are properly accessed before the models are created and applied.
• Evaluate the model without overevaluating. The model must be tested to ensure that it provides better decision-making capabilities over current analysis methods.
• Share the results. The insights that result from the predictive analysis must be shared with everyone across the district, including teachers, counselors and parents.