Machine Learning with Immersion

Immersion is a prediction engine. It was designed from the get-go to accurately and consistently predict what individuals and markets will do. This linkage means that clients can measure the value Immersion creates immediately and unambiguously.  

Immersion's scientists have shown that the metrics from the platform consistently predict market outcomes with 80% or greater accuracy. Many of these models have been published in peer-reviewed scientific journals. Some of these models were simple and some quite sophisticated. Our team continues to build statistical models to improve Immersion's predictive accuracy.

We recently shared how Immersion's metrics complement traditional measures used by market researchers. Neurologic data are a statistically independent source of information about customers and thus improve the accuracy of traditional predictive models. We never said that neurologic data were the only thing one should measure.

Until now.

Our collaboration with Pandora showed that Immersion could identify new music hits with 83% accuracy using a standard statistical approach. Let's be honest, 83% is remarkably accurate since humans reporting if they "like" a song predicts hits with 0% accuracy. Then we started to wonder if we could we beat 83%.

OK, time to get your nerd on. We investigated several machine learning approaches by creating large synthetic datasets from the original data. One-half of the data was used to train several different machine learning models and predictions were made on the other half. We then compared each model to see if we could exceed the 83% accuracy hurdle.

Using the two Immersion metrics from the original Pandora analysis, a k-nearest neighbor model identified hits with 97.2% accuracy. This predictive accuracy was highly statistically significant.   Among the 24,000 new songs released every day, Immersion used alone can almost perfectly identify the roughly 400 that will be hits months in advance of their release. Record producers, radio stations, and streaming services can delight listeners by playing these 400 songs and mostly avoiding the other 23,600.

The return on investment is obvious: producers focus on supporting the best music, more people listen to this music and commercials on the radio, and more people subscribe to streaming services. All because music-lovers' brains are immersed.

Immersion's clients use the platform as workflow software to optimize content creation and delivery to ensure that what they release is a hit. Our clients have shown that what works for music also predicts hits for TV, movies, training, education, customer journeys, live events and more.

Find out how easy it is to harness the power of Distributed Neuroscience^TM at Immersion.