Uber. Siri. Spam filters. Netflix. iPhoto. Mobile check deposits. Google. Spotify. Even some medical diagnoses. What do all of these things have in common? They utilize revolutionary computer insight to predict the future. It seems like science fiction, but computer insight, otherwise known as artificial intelligence (AI), is becoming a commonplace fixture of daily life in the 21st century. It’s ubiquitous, making everyday tasks simpler and providing a convenience only a few ever thought possible. The specific algorithms that allow these modern-day advancements often make use of machine learning (ML) techniques. According to ML pioneer Arthur Samuel, “[machine learning is the] field of study that gives computers the ability to learn without being explicitly programmed.” It is one of many ways to effectuate artificial intelligence, and it makes tasks easier for humans.
In the world of market research, machine learning is providing increasingly advantageous ways to target consumers, sort data, predict product performance and customer behavior, place products in the marketplace and drive results. This is especially important because researchers are gathering more and more data. In fact, so much data is being generated that strict human interpretation is becoming impossible. Machine learning can manipulate these big data sets by storing, analyzing and applying fresh and more relevant parameters each time it encounters new information. It learns something with each data cycle and becomes better and better at predicting meaningful results. For a person to accomplish the same would be not only impractical, but probably impossible, as well.
There are basically three ways to use machine learning:
Supervised Learning – Such as Classification and Regression
With supervised learning, machines are equipped with a sample data set that it then uses it to predict responses within a new data set. Qualitative (classification) or numeric (regression) data can be calculated and provides market researchers with opportunities to group segments by similar characteristics. For instance, companies might use an algorithm to answer the question: “Is this segment likely or not likely to purchase a product over $10?” By grouping historical data into groups based on consumer/non-consumer type and their location, income level or other variables, you can train the computer to pinpoint the characteristics that best represent a customer willing to pay $10 or more for a product. It can then analyze a new data set for similar characteristics, leading you to labeling old or identifying new customers. A regression model uses numeric data instead of tags, but functions the same way.
Unsupervised Learning – Such as Clustering
Unsupervised learning means the machine determines data patterns on its own without initial input. It basically clusters a data set into groups based on similarities. Marketers might use it to determine if people purchasing a specific product have a similarity that can be identified and then targeted in the future.
In reinforcement learning, machines learn through “trial and error” or “cause and effect.” It’s a mixture of supervised and unsupervised learning and uses positive and negative feedback as reinforcement for future predictions. In essence, the computer learns to respond independently to the environment based on previous encounters. Market researchers can use reinforcement learning to recommend products to consumers, much like Netflix does when it recommends those movies for you to watch based on your previous viewing habits and ratings.
In short, machine learning is a great tool for businesses wanting to expand the breadth of information available from their data sources. If you’re interested in learning if it can help you garner actionable results, contact us at Communications for Research (CFR). Our co-CEO Colson Steber can review your market research objectives and determine if machine learning can be applied to your situation.
For more information on how to communicate the value of market research to your clients, download our FREE eBook: