What are the most troubling pitfalls of analytics? Here’s our countdown of the top 5 mistakes to avoid when using predictive analytics.
5. Thinking Too Big or Too Small
While your overall analytics goal may be to optimize profitability, it is important to predict the individual components that drive profitability (such as response and conversion). On the other hand, you do want to have some faith in the power of analytics, which can be used to predict medium sized things that add up to big business impacts.
4. Predicting The Past Instead of The Future
It is common to see models that perfectly predict what did happen. Modeling techniques that extensively use historical data may seem great, but usually do not do a good job of predicting the future.
3. Making The Assumption That Analytics is a Pure Science
While ideal analysts often have a technical academic background, there’s actually an art to building effective statistical models. More important is an understanding of the business issues driving the need and then leveraging statistical tools to achieve the goals.
2. Data Bias
It’s very important to spend time evaluating the input data prior to executing the analysis. Here’s an important formula:
(bad data in)=(bad solution out)
1. Predicting The Wrong Thing
Real experience is needed to understand what dependent variable data is needed to reach a particular business goal. Pragmatism is almost always vital to success in the real world of messy data and fuzzy goals.