When I first saw the letters AI and ML I was a bit dumbfounded. I’m a fairly technologically competent guy, but for a while these were terms placed on a titanium throne with encrusted sapphires rarely discussed as passing conversation. Fast Forward 13 years, and you can walk down the sidewalk and hear the terms come up more often than not — especially if you live in Boston.

As humans we have a tendency to point to the next big thing. What is the next innovation? How will this make my life easier? What’s in it for me? All of these questions typically conclude with “well damn that’s pretty neat.” The same may be stated about the use of machine learning for recommendation engines from restaurant sales and marketing data. To fully realize the potential of deep data insights provided by machine learning, we will explore a succinct, yet broad, definition of machine learning:

Machine Learning involves the use and creation of learning models in which a system has the ability to learn and improve without being explicitly programmed.

The definition above encourages us to think about how machine learning is relevant to the restaurant point-of-sale system. What are the common questions restaurant owners ask themselves on a daily basis in regard to their restaurant and how can we qualify and quantify those questions to recommend action(s) and insight(s)? Enter the great superhero of our day ML (Machine Learning).

Well, first we need the data; something the program can learn from, react to, process and analyze. Alongside the data we need the algorithms to deeply consider what we need accomplished presently and in the future. See what we want is for the program to evolve and gain more accuracy as more data is provided. With the accuracy intact, the algorithms may then start to increase output of valuable insights and recommendations.

So how does a program learn?

Well to put a long and detailed story short, there are generally 3 recognized learning types:

Supervised Learning(SL)

Supervised Learning has an input and an output searching for a specific pattern within the data from a given set of predictors. You can think of output in this case the dependent variable, and the predictors as independent variables.

Unsupervised Learning(UL)

Unsupervised learning algorithms do not have a target output, rather, they are used to predict such things as anomalies and patterns.

Reinforcement Learning(RL)

Possibly the pinnacle of artificial intelligence where in a program learns similar to the way we do clicker training with dogs. Furthermore, reinforcement learning enables a machine to to learn behavior based on environment, scenario, and actions in which the correct actions will be rewarded.

Each learning type produces a different outcome which may provide the information sought out. Who knows, maybe Reinforcement Learning would be the way to go with a POS recommendation system, or perhaps a hybrid learning type — i’m not an expert, I just enjoy thinking about these things.

So what’s the point?

Right now, our POS systems provide a fair amount of data and reporting tools — especially with the onset of cloud-based platforms. But, do you see the problem here? These are only reports with good visuals that provide actionable data, but does not fully provide the insight needed.

The sales and marketing data collected and used in reports by a restaurant POS recommendation system might be able to push past the limitation of decision making by realizing possible outcomes and approaches.

No, the idea is not very far-fetched. In the mPOS community, which is primarily iPad-Based, Apple’s strong interest in Augmented Reality and Machine Learning makes it possible for mPOS developers to sandbox machine learning elements and ideas for their relative companies. Machine Learning does not stop at providing recommendations to managers and business owners. Imagine a point in which, through Natural Language Processing a POS can pick up on order items through a server or manager’s voice.

I will not say that the possibilities are endless because endless is incomprehensible, but the engagement and creativity developed go a long way to re-designing not only the customer experience, but that of the restaurateur.