JOUR 651 Lecture Notes - Lecture 31: Machine Learning
Document Summary
There are many things that can be behind estimated metrics, such as: Once you understand this, you can put the right metrics needed to make good use of them for your strategy. Two common applications of estimated metrics: correlation and patterns between true performance and estimates, supporting metrics: from substitutions of missing data, to analytics that are prescriptive and/or predictive in nature. A common thing to do when using estimated metrics is looking for correlations between: We need to ask ourselves if when we have higher reach, if that means having more interactions. The thing about this is that it"s hard to judge what a good end number should be, so if you are comparing reach and impressions, don"t focus on the final number to indicate performance. To conduct this kind of analysis, you need to check into the correlations of the metrics you"ve built.