Gartner has published the Hype Cycle for Advanced Analytics and Data Science 2015 chart in July 2015. It shows that uplift modeling is an innovation trigger now and is going to achieve "Plateau of Productivity" stage during 5 to 10 years.
When we were starting research in uplift modeling over twelve years ago, we strongly believed that this solution has the great potential. The first commercial implementation of uplift modeling in 2004 only confirmed our conviction. Since then, we have developed many machine learning algorithms that allow to create uplift models.
Part of our widely known algorithms, such as target variable transformation for uplift modeling and the first version of uplift decision trees adapted to build random forests are used all over the world, for examples in R uplift package. Moreover the most popular machine learning algorithms, like logistic regression, decision trees, boosting, support vector machine and survival analysis we have adapted to uplift modeling and have published in scientific journals.
New algorithms and the newest versions of those commonly known have been tested in real world cases, and are only used in our Decitum Action Optimizer solution.
We are very proud having so large contribution to uplift modeling. Hope to see you on that innovation bandwagon soon.