What is AI & what Can it Do for My Business?
This is the second installment of our blog series of Data Science and Artificial Intelligence posts by Dr. Colin O’Callaghan of 3Advance. If you missed part one you can read it here.
Recently, there have been huge jumps in data science techniques for automatic image, language, audio, and data processing. This has accelerated many technologies that are having an increasing impact on all industries, which will add trillions to global GDP over the next decade according to PwC.
As with any technology, new concepts and terminology are being used to talk about data science and Artificial Intelligence (AI). In this post, we will define the most important of these phrases: data science, machine learning, and deep learning. We will also give some examples of how these concepts are being used to revolutionize different industries.
Data science combines statistics and computer science to extract knowledge from large and complex data-sets. With the increasing digitization of all industries, data science has become more relevant and useful to companies in recent years. By combining industry expertise with data science techniques, companies have leveraged their data to identify patterns that they can act on.
When reading about data science, you are going to come across the phrases machine learning, deep learning, and AI. These phrases usually bring to mind an intelligent robot like C-3PO or Wall-E. To be clear, AI technology is nowhere near that advanced! AI is defined as simulating intelligent behavior in machines, or in other words, decision-making in computers. This means that AI software needs to be able to identify an intelligent response to a set of inputs.
Now, people have been developing software to do this since the first computers. Programmers define different conditions that trigger certain responses in software. Traditionally, it was programmers who decide what the conditions are and the desired response in each case. What if a computer could figure these conditions and responses out for themselves? Enter machine learning.
Machine learning is a broad set of algorithms that build predictive models without having to explicitly program relationships and analysis rules in your data. To do this, the machine learning algorithms need a large, high quality data-set from which it can “learn” patterns.
Let’s use a modern email spam detector as an example. Consider a data-set of emails that includes the sender address, subject line, email body, type of file attachments, etc. This data is the input to a machine learning algorithm. Some algorithms will also need target outputs, in this case whether an email is spam or not. The machine learning algorithm then “learns” from this data, determining what combinations of inputs makes an email likely to be spam or not. A key takeaway is that no matter how good the machine learning algorithms are, they are limited by the volume and quality of the training data.
Another term regularly seen in articles about AI is deep learning. This is a subset of machine learning, and it is suited to very complex data. Some of the more common applications of deep learning are in speech recognition, language analysis, and image processing (often referred to as computer vision). Have you used Siri, Google Translate, or the facial recognition feature on your camera recently? Deep learning was used to develop those technologies. Deep learning is layers of several machine learning and data processing algorithms that feed processed data into one another before producing a final set of outputs. Essentially, deep learning is a more powerful machine learning.
Examples of Data Science for Businesses
Let’s look at some examples of how some of these techniques are being used in different industries.
Recommender systems are used by many different services — such as online retail and streaming services — to suggest new content or products they think you may like. For example, you watched The Lord of the Rings? You might like Game of Thrones! Well, seasons 1–7 anyway. How does this work? Well, these services have huge databases of user consumption history. By trawling through these databases and applying machine learning techniques, algorithms can identify patterns in user preferences. This means a more personalized experience for the users and more sales or higher user ratings for the companies themselves.
When it comes to data impacting the sports industry, one book captures this best, Moneyball. Michael Lewis’s book and its movie adaptation showed the incredible success that early adopters of data science can achieve. Moneyball follows the MLB’s Oakland A’s and how data science was used to guide their recruitment strategy, leading to more wins for less money back in 2002.
Data science is taking off in sports that until recently were seen as too complex for statistical analysis. For example, a recent article in the New York Times lifted the lid on Liverpool FC’s data science team who had been advising management since 2015 and helped them win the champion’s league for the first time since 2005. Back in the US, many NBA teams are using data science to decrease injuries, increase scores, improve strategy, and more. From automatic video analysis with deep learning to better training regimes, the sports industry is being revolutionized by data science. So much potential exists in the intersection of data and sports that it will be the subject of more of our blogs in the future, so subscribe to our newsletter to learn more!
Another common way data science can be used for businesses is to improve sales. You probably have a huge database of customer data stored on different web-services. This data can be analyzed to shed light on the characteristics of your customers, such as what products are they are attracted to, their purchasing habits, and the marketing strategies likely to influence them. Data science can be used to quantify all of this, allowing you to hone your marketing and to get your products into the hands of those who want it. A win-win.
A great use case of data science is the ability to scan through huge volumes of text and extract human-level insights. For example, many social media platforms will use machine learning techniques to identify bots and trolls. Algorithms can identify the emotion associated with sentences and words, giving unprecedented insights into social media posts, news and blog posts, customer reviews, and even customer service responses. Many automated chatbots also make use of this kind of technology. Google Translate is another great example of machine learning for language analysis, making use of deep learning to create accurate translations automatically.
Data Science in Your Business
So you might be wondering, what can data science do for me? Well, it depends. Do you have a large database with predictive power you want to leverage? Our data science consulting team can help you identify solutions. Do you have large amounts of image, video, audio, or text data that you want to extract information from? Our machine learning consulting team can implement software packages that have been developed with deep learning models to extract data from it automatically. Not sure where to start with all of your different data sources and what questions to ask of it? Talk to our big data team for ideas or read our recent blogs about projects we are working on and taking your first steps in gathering, storing, and analyzing your data.
3Advance is an app development company in Washington DC that creates beautiful, simple mobile apps. Get in touch with our data science consulting team if you’d like to learn more about how machine learning, deep learning and AI can move your business forward.