As Told By “Her”: The Power of Predictive Analytics


Share on LinkedIn

Is it possible our phones could one day know us better than our best friend, spouse … and even ourselves?

As seen in the recent Oscar-winning film “Her,” a man named Theo develops a meaningful (and then romantic) relationship with an artificial intelligence-infused computer operating system. The OS, named Samantha, draws on a myriad of data to learn Theo’s likes, dislikes, desires, and predictable behaviors—exploring the capacity machines have to understand, predict, adapt to, and interact with humans.

“Her” may be set in the future, but we’re probably closer to this reality than we realize. The heart of this ability, often referred to as machine learning or predictive analytics, lies in the analysis of data to pinpoint indicators that lead to concrete outcomes. Pair all the data on your phone with a powerful big data analytics tool, and a world of discoveries will be amazingly, and unnervingly, revealed.

What Your Phone Knows

Predictive analytics can use data to analyze how you feel, what you do, when you do it, what you like, why you like it, and more. Here are a few examples of rich data your phone would synthesize to better understand you:

  • Email and text messages: Your phone can analyze when you write, who you write to, who writes to you, how many messages you receive, what’s on your mind, and the sentiment of these messages.
  • Third-party apps: The types of apps you use (productivity, entertainment, games, etc.) and the data within them (e.g., Instagram photos…).
  • Social media: A rich source of data, social media provides photos, posts, likes, friends, messages, and more.
  • Photos and videos: Beyond what’s in the photo itself, your phone can track time, location, frequency, friends, etc.
  • Calendar: Your calendar reveals frequency and types of events, hours spent on specific activities, who you’re doing things with, etc.
  • Browser: Data from the websites you visit, words you search, and all content associated with these sites is extremely useful for predictive analytics.
  • Music: In “Her,” Theo commands his phone to “Play a melancholy song,” and his OS knows exactly what he’s in the mood for.
  • Weather: Most people don’t realize how much the weather affects their thoughts, feelings and actions. But an OS can easily detect these types of connections.

Relating data is where computation really shines beyond human ability. Your phone understands not only what you do, but relates it to others. It learns that people like to play upbeat music on weekend evenings. And when you just Googled “fun weekend activities” and looked up a comedy on Netflix, your phone has learned that others who do this are in good moods, so it also recommends that you listen to an upbeat song that others listen to when they’re happy. Machine learning can also identify surprising correlations in human behavior.

Predictive Power is Everywhere

Eric Siegel’s recent book, Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die, identifies the following examples: When people buy a Barbie doll, they’re also likely to buy one of three brands of candy bars; vegetarians miss fewer airline flights; and you’re likely to be happier when Facebook friends post positive messages.

In “Her,” Samantha compiles all of Theo’s best work into a book and sends it to a publisher because she correctly predicts this is a dream Theo has been too scared to carry out. In real life, if you were Theo, your phone could see that you research how to invest better, visit job sites, filter e-commerce product lists by cost, and visit book publisher sites frequently. From this, an OS senses financial and professional discontent, as well as aspirations. And it knows through some texts and emails that you think your professional work is really good.

Predictive analytics is all around us. When Netflix recommends a movie, it now relies less on how you rated other movies and more on what you browse and what you watch, because your actual behavior leads to better recommendations. Amazon is even experimenting with delivering specific items to a nearby warehouse before you buy because it predicts you will do so shortly. And the IRS forecasts whether you’re likely to cheat on your taxes, while HP predicts if an employee is going to quit.

For a few years now, Samsung has been developing technology that predicts user emotions. Imagine your phone choosing a ringtone that communicates the caller’s mood. Samsung can also monitor moods by how fast you type when texting, how much your phone shakes in your hands, and how often you backspace when typing. Consequently, your phone could then show you a cartoon to make you laugh if it senses you’re unhappy.

How Will You Tap the Power in Big Data?

Considering the power that lies in big data, perhaps it’s not so impossible for a computer to know so much about people that it learns to truly understand them, better than they even understand themselves. In this way, “Her” gives us a poignant glimpse into the future. The possibilities of machine learning are vast and exciting, and the potential to create a predictive framework for individuals, businesses and even societies carries significant implications for improving quality of life.

You may not fall in love with your phone, but get ready for a closer relationship, because no one may know you better than it does.

Joseph Pigato
Joseph Pigato is the Managing Director of Sparked, which helps companies retain their customers through sophisticated predictive analytics and engagement tools. Follow Sparked on Twitter @sparked.


  1. It seems we’re heading for a future where data generated unknowingly by devices on our behalf (like location data) is a greatly in demand product. One question we need to answer is that of ownership, as well.

  2. Wow, imagine what our gadgets that we have today will do in the future! Thanks for all the extra information! This has really changed my view on technology. Thanks for the great post!

  3. Great post as usual, the reality is that most human things will be digitized and people have a million data points so it will take a bit for the analysis to be valuable, or it will be like economics and it will be theory mixed with predictability. Either way for analyzing things it is just going to grow and get more effective. Sharing :]

  4. Agreed – there’s power in predictive analytics. Great power. But there’s great danger that accompanies boundless enthusiasm for that power. Nate Silver wrote in “The Signal and the Noise” that some of the biggest problems with predictive analytics occur when there is too much credence given to the output. In other words, pretty bad things can happen when people over-estimate the accuracy of their predictions. I understand this. We all love the cool things algorithms can do, especially when we formulate them ourselves.

    I see this problem frequently when companies amass mountainous terabytes of data, and seek to find uses for it all to predict – well, almost anything. Another issue that’s often brought up is that predictive analytics can be extraordinarily valuable at answering “what” and “when, but “why” is rarely, if ever, answered. I scoured this blog, but didn’t find this question addressed.

    This issue is much debated. Whereas traditional market research has sought to discover why consumers behave in a certain way, modern analytics bypasses the causal factors in favor of knowing, for example, that a certain user prefers comedies over all other film genres, but selects dramas only during the winter. “We don’t know why – it is what it is, so that’s what we promote in the search results.”

    Your phone “learns that people like to play upbeat music on weekend evenings,” but who is developing insight into why this occurs? Great as our nascent predictive capabilities are, I’m not sure about the cost of trade-offs being made regarding innovation. Without deeply exploring “why”, can companies develop what people want or need, or deliver sustainable value to their customers?


Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here