<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Monica Rogati on Medium]]></title>
        <description><![CDATA[Stories by Monica Rogati on Medium]]></description>
        <link>https://medium.com/@mrogati?source=rss-b31aba7f4828------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dpJkmxtq8NurSHpFr4LRUA.jpeg</url>
            <title>Stories by Monica Rogati on Medium</title>
            <link>https://medium.com/@mrogati?source=rss-b31aba7f4828------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 06:47:15 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mrogati/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com" rel="hub"/>
        <item>
            <title><![CDATA[The AI Hierarchy of Needs]]></title>
            <link>https://medium.com/hackernoon/the-ai-hierarchy-of-needs-18f111fcc007?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/18f111fcc007</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Tue, 01 Aug 2017 14:05:04 GMT</pubDate>
            <atom:updated>2019-05-22T09:19:55.097Z</atom:updated>
            <content:encoded><![CDATA[<p>As is usually the case with fast-advancing technologies, AI has inspired massive <a href="https://en.wikipedia.org/wiki/Fear_of_missing_out">FOMO</a> , <a href="https://en.wikipedia.org/wiki/Fear,_uncertainty_and_doubt">FUD</a> and <a href="https://techcrunch.com/2017/07/25/elon-musk-mark-zuckerberg-artificial-intelligence/">feuds</a>. Some of it is deserved, <a href="https://chatbotnewsdaily.com/machine-learning-ai-and-the-emperors-vest-32b7bdd99b58">some of it not</a> — but the industry is paying attention. From stealth hardware startups to fintech giants to public institutions, teams are feverishly working on their AI strategy. It all comes down to one crucial, high-stakes question: <strong><em>‘How do we use AI and machine learning to get better at what we do?’</em></strong></p><p>More often than not, companies are <strong>not</strong> ready for AI. Maybe they hired their <a href="https://hackernoon.com/how-not-to-hire-your-first-data-scientist-34f0f56f81ae">first data scientist</a> to less-than-stellar outcomes, or maybe <a href="https://artplusmarketing.com/data-literacy-product-design-and-the-many-faced-god-cf8339e035a">data literacy</a> is not central to their culture. But the most common scenario is that they have not yet built the infrastructure to implement (and reap the benefits of) the most basic <a href="https://hackernoon.com/tagged/data-science">data science</a> algorithms and operations, much less <a href="https://hackernoon.com/tagged/machine-learning">machine learning</a>.</p><p>As a data science/AI advisor, I had to deliver this message countless times, especially over the past two years. Others <a href="https://hbr.org/2017/06/if-your-company-isnt-good-at-analytics-its-not-ready-for-ai">agree</a>. It’s hard to be a wet blanket among all this excitement around your own field, especially if you share that excitement. And how do you tell companies they’re not ready for AI without sounding (or being) elitist — a self-appointed gate keeper?</p><p>Here’s an explanation that resonated the most:</p><p><strong>Think of AI as the top of a </strong><a href="https://www.simplypsychology.org/maslow.html"><strong>pyramid of needs</strong></a><strong>. Yes, self-actualization (AI) is great, but you first need food, water and shelter (data literacy, collection and infrastructure).</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7IMev5xslc9FLxr9hHhpFw.png" /><figcaption>You need a solid foundation for your data before being effective with AI and machine learning.</figcaption></figure><h4>Basic needs: Can you count?</h4><p>At the bottom of the pyramid we have <strong>data collection</strong>. What data do you need, and what’s available? If it’s a user-facing product, are you logging all relevant user interactions? If it’s a sensor, what data is coming through and how? How easy is it to log an interaction that is not instrumented yet? After all, the right <a href="https://qz.com/1034972/the-data-that-changed-the-direction-of-ai-research-and-possibly-the-world/">dataset</a> is what made recent advances in machine learning possible.</p><p>Next, how does the <strong>data flow</strong> through the system? Do you have reliable streams / ETL ? Where do you store it, and how easy is it to access and analyze? <a href="https://twitter.com/jaykreps">Jay Kreps</a> has been saying (for about a decade) that reliable data flow is key to doing anything with data. <em>[Aside: I was looking for an exact quote and found it in his ‘</em><a href="https://engineering.linkedin.com/distributed-systems/log-what-every-software-engineer-should-know-about-real-time-datas-unifying"><em>I love logs</em></a><em>’ masterpiece. I then noticed that, one paragraph over, he’s making this exact Maslow’s hierarchy of needs comparison, with an ‘it’s worth noting the obvious’ thrown in there for good measure (thanks Jay!). Speaking of related work, I’ve also later run (h/t Daniel Tunkelang) into Hilary Mason and Chris Wiggins’s excellent </em><a href="http://www.dataists.com/2010/09/a-taxonomy-of-data-science/"><em>post</em></a><em> about what a data scientist does. Days ago, Sean Taylor </em><a href="https://twitter.com/mrogati/status/888120281276891136"><em>unveiled</em></a><em> his own data science pyramid of needs (ironically dubbed the Unconjoined Triangle of Data Science) which, of course, is completely different. Perhaps we should start a tumblr.]</em></p><p>Only when data is accessible, you can <strong>explore and transform</strong> it. This includes the infamous ‘data cleaning’, an under-rated side of data science that will be the subject of another post. This is when you discover you’re missing a bunch of data, your sensors are unreliable, a version change meant your events are dropped, you’re misinterpreting a flag — and you go back to making sure the base of the pyramid is solid.</p><p>When you’re able to reliably explore and clean the data, you can start building what’s traditionally thought of as BI or <strong>analytics</strong>: define metrics to track, their seasonality and sensitivity to various factors. Maybe doing some rough user segmentation and see if anything jumps out. However, since your goal is AI, you are now building what you’ll later think of as <strong>features</strong> to incorporate in your machine learning model. At this stage, you also know what you’d like to predict or learn, and you can start preparing your <strong>training data</strong> by generating labels, either automatically (which customers churned?) or with humans in the loop.</p><p>This is also when you find your most exciting and compelling <strong>data stories</strong> — but that’s also the subject of another Medium post.</p><h4>OK, I can count. Now what?</h4><p>We have training data — surely, now we can do machine learning? Maybe, if you’re trying to internally predict churn; no, if the result is going to be customer-facing. We need to have a (however primitive) A/B testing or <strong>experimentation </strong>framework in place, so we can deploy incrementally to avoid disasters and get a rough estimate of the effects of the changes before they affect everybody. This is also the right time to put a very <strong>simple baseline</strong> in place (for recommender systems, this would be e.g. ‘most popular’, then ‘most popular for your user segment’ — the very annoying but effective ‘stereotype before personalization’).</p><p><em>Simple heuristics are surprisingly hard to beat, and they will allow you to debug the system end-to-end without mysterious ML black boxes with hypertuned hyperparameters in the middle. This is also why my favorite data science algorithm is division.</em></p><p>At this point, you can deploy a very simple ML algorithm (like logistic regression or, yes, division), then think of new signals and features that might affect your results. Weather &amp; census data are my go-tos. And no — as powerful as it is, deep learning doesn’t automatically do this for you. Bringing in new signals (feature creation, not feature engineering) is what can improve your performance by leaps and bounds. It’s worth spending some time here, even if as data scientists we’re excited about moving on to the next level in the pyramid.</p><h4>Bring on the AI!</h4><p>You made it. You’re instrumented. Your ETL is humming. Your data is organized &amp; cleaned. You have dashboards, labels and good features. You’re measuring the right things. You can experiment daily. You have a baseline algorithm that’s debugged end-to-end and is running in production — and you’ve changed it a dozen times. You’re ready. Go ahead and try all the latest and greatest out there — from rolling your own to using companies that specialize in machine learning. You might get some big improvements in production, or you might not. Worst case, you learn new methods, develop opinions and hands-on experience with them, and get to tell your investors and clients about your AI efforts without feeling like an impostor. Best case, you make a huge difference to your users, clients and your company — a true machine learning success story.</p><h4>Wait, what about MVPs, agile, lean and all that?</h4><p><strong>The data science hierarchy of needs is not an excuse to build disconnected, over-engineered infrastructure for a year.</strong> Just like when building a traditional MVP (minimally viable product), you start with a small, vertical section of your product and you make it work well end-to-end. You can build its pyramid, then grow it horizontally. For example, at Jawbone, we started with sleep data and built its pyramid: instrumentation, ETL, cleaning &amp; organization, label capturing and definitions, metrics (what’s the average # of hours people sleep every night? What about naps? What’s a nap?), cross-segment analyses all the way to <a href="https://jawbone.com/blog/jawbone-up-data-by-city/">data stories</a> and machine learning-driven data products (automatic sleep detection). We later <a href="https://jawbone.com/blog/best-health-fitness-data-stories-2015/">extended</a> this to steps, then food, weather, workouts, social network &amp; communication — one at a time. We did not build an all-encompassing infrastructure without ever putting it to work end-to-end.</p><h4>Asking the right questions and building the right products</h4><p>This is only about how you <strong>could</strong>, not whether you <strong>should</strong> (for pragmatic or ethical reasons).</p><h4>The promise of machine learning tools</h4><p>‘<em>Wait, what about that Amazon API or TensorFlow or that other open source library? What about companies that are selling ML tools, or that automatically extract insights and features?</em>’</p><p>All of that is awesome and very useful. (Some companies do end up painstakingly custom-building your entire pyramid so they can showcase their work. They are heroes.) However, under the strong influence of the current AI hype, people try to plug in data that’s dirty &amp; full of gaps, that spans years while changing in format and meaning, that’s not understood yet, that’s structured in ways that don’t make sense, and expect those tools to magically handle it. And maybe some day soon that will be the case; I see &amp; applaud efforts in that direction. Until then, it’s worth building a solid foundation for your AI pyramid of needs.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2Fdde502%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2Fhackernoon%2F&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/3c851dac986ab6dbb2d1aaa91205a8eb/href">https://medium.com/media/3c851dac986ab6dbb2d1aaa91205a8eb/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18f111fcc007" width="1" height="1" alt=""><hr><p><a href="https://medium.com/hackernoon/the-ai-hierarchy-of-needs-18f111fcc007">The AI Hierarchy of Needs</a> was originally published in <a href="https://medium.com/hackernoon">HackerNoon.com</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How not to hire your first data scientist]]></title>
            <link>https://medium.com/hackernoon/how-not-to-hire-your-first-data-scientist-34f0f56f81ae?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/34f0f56f81ae</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[big-data]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Tue, 07 Mar 2017 16:05:58 GMT</pubDate>
            <atom:updated>2019-05-22T09:19:54.641Z</atom:updated>
            <content:encoded><![CDATA[<p>You’re a <a href="https://hackernoon.com/tagged/startup">startup</a> founder. You know you need to have a “data play” (or worse, “AI play”). Investors and clients are asking about machine <a href="https://hackernoon.com/tagged/learning">learning</a> (or worse, deep learning). The question is no longer why, but when.</p><p>So, you hire your first data scientist. A fresh graduate (those with industry experience are too expensive and hard to recruit). They have a PhD (so you can nonchalantly drop ‘I have my best PhDs looking into this’ in the conversation with potential clients and VCs.) They‘re excited about doing machine learning — and so are you. This is going to rock!</p><p>Your data scientist starts, ready to take your data and build models from it, just like you agreed. Back in academia, they were working with datasets that several generations of grad students have been using. They’re not naive, though — they know this time around they’ll have to create their own. No problem! Your startup has a year worth of logs in S3, ready to be mined and used as training data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oxzPOhyANmEiNwnhrGz5Wg.png" /><figcaption>How not to think about the data science Venn diagram</figcaption></figure><p>There’s no infrastructure for data analysis, so everything has to be done from scratch. Your data scientist is trying to install various tools incompatible with your stack or architecture and, for now, is doing one-off log parsing to answer every question. The engineering team doesn’t feel it’s safe to give them prod access so they provide an offline copy of your database.</p><p><em>There!</em> you say to the data scientist. <em>You have access to all the data you need!</em></p><p>Except… data in that offline database is, by design, not structured in a way that’s easy to analyze or combine with the data in the logs. Values don’t make sense or are missing for weeks at a time, the only events instrumented are those relevant to the ops team, and even the simplest query takes forever to run.</p><h4>Everybody is frustrated .</h4><p>The data scientist wanted to work on machine learning — those elegant algorithms they spent their career thinking and publishing about. Yes, they expected to put some effort into gathering and cleaning data — but they didn’t expect it to be so complex and messy, with so much missing or hard to access. They didn’t expect to spend so much time in meetings or Slack with questions about how the data was gathered and what, exactly, is in that json blob. They didn’t expect the rest of the company to care so little about how each redesign messes up month to month comparisons. They didn’t realize nobody logged all the options a user saw before they clicked, nor the version of the UI at the time, so that one year of training data is not exactly usable.</p><p>You hired a data scientist to work on machine learning. They’ve been warned about spending 80% of their time cleaning data. That sounds like a distant dream now. Instead, they spend 80% of their time begging for data to be created, accessed, moved, or explained. They’re spending the other 20% lobbying for data science-friendly tools, security policies, and infrastructure, or looking for a new job.</p><p>The engineering team is frustrated, too. They have to take time away from their own work to do a ton of thankless grunt work for the scientist who sits around, does nothing useful and still manages to constantly complain about the data not being good enough.</p><p>You, the founder, are even more frustrated. It’s been 2 months, and the data scientist didn’t even produce a decent dashboard, much less the magic machine learning everybody keeps raving about. Plus, they seem like a bad culture fit, slowing everybody else down for no reason. Deep down, you start suspecting machine learning is snake oil and <a href="https://chatbotnewsdaily.com/machine-learning-ai-and-the-emperors-vest-32b7bdd99b58#.lomvxohgc">AI is just hype</a>.</p><p>You keep this to yourself, though, and tell your investors and clients about your AI play and your crazy smart PhD data scientist who’s super excited to work on machine learning any day now.</p><p><em>If you’re wondering how to avoid this painful but too frequent scenario, here’s the Quora answer that inspired this post : “</em><a href="https://www.quora.com/What-are-the-challenges-of-building-a-data-team-at-a-startup"><em>What are the challenges of building a data team at a startup</em></a><em>?”.</em></p><h3>Think other people would benefit from this article? Click the heart to help them find it ❤</h3><figure><a href="http://bit.ly/HackernoonFB"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0hqOaABQ7XGPT-OYNgiUBg.png" /></a></figure><figure><a href="https://goo.gl/k7XYbx"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Vgw1jkA6hgnvwzTsfMlnpg.png" /></a></figure><figure><a href="https://goo.gl/4ofytp"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gKBpq1ruUi0FVK2UM_I4tQ.png" /></a></figure><blockquote><a href="http://bit.ly/Hackernoon">Hacker Noon</a> is how hackers start their afternoons. We’re a part of the <a href="http://bit.ly/atAMIatAMI">@AMI</a> family. We are now <a href="http://bit.ly/hackernoonsubmission">accepting submissions</a> and happy to <a href="mailto:partners@amipublications.com">discuss advertising &amp; sponsorship</a> opportunities.</blockquote><blockquote>If you enjoyed this story, we recommend reading our <a href="http://bit.ly/hackernoonlatestt">latest tech stories</a> and <a href="https://hackernoon.com/trending">trending tech stories</a>. Until next time, don’t take the realities of the world for granted!</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*35tCjoPcvq6LbB3I6Wegqw.jpeg" /></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2Fdde502%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2Fhackernoon%2F&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/3c851dac986ab6dbb2d1aaa91205a8eb/href">https://medium.com/media/3c851dac986ab6dbb2d1aaa91205a8eb/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=34f0f56f81ae" width="1" height="1" alt=""><hr><p><a href="https://medium.com/hackernoon/how-not-to-hire-your-first-data-scientist-34f0f56f81ae">How not to hire your first data scientist</a> was originally published in <a href="https://medium.com/hackernoon">HackerNoon.com</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How do I become a data scientist?]]></title>
            <link>https://medium.com/good-audience/how-do-i-become-a-data-scientist-f8074232608e?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/f8074232608e</guid>
            <category><![CDATA[education]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Mon, 13 Feb 2017 21:48:11 GMT</pubDate>
            <atom:updated>2018-08-29T22:29:38.699Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>tl;dr: Do a project you care about. Make it good and share it.</strong></p><p>There’s a lot of interest in becoming a data scientist, and for good reasons: high impact, high job satisfaction, high salaries, high demand. A quick search yields a plethora of possible resources that could help — MOOCs, blogs, Quora answers to this exact question, books, Master’s programs, bootcamps, self-directed curricula, articles, forums and podcasts. Their quality is highly variable; some are excellent resources and programs, some are click-bait laundry lists. Since this is a relatively new role and there’s no universal agreement on what a data scientist does, it’s difficult for a beginner to know where to start, and it’s easy to get overwhelmed.</p><p>Many of these resources follow a common pattern: 1) here are the skills you need and 2) here is where you learn each of these. Learn Python from this link, R from this one; take a machine learning class and “brush up” on your linear algebra. Download the iris data set and train a classifier (“learn by doing!”). Install Spark and Hadoop. Don’t forget about deep learning — work your way through the TensorFlow tutorial (the one for ML beginners, so you can feel even worse about not understanding it). Buy that old orange Pattern Classification book to display on your desk after you gave up two chapters in.</p><p>This makes sense; our educational institutions trained us to think that’s how you learn things. It might eventually work, too — but it’s a unnecessarily inefficient process. Some programs have capstone projects (often using curated, clean data sets with a clear purpose, which sounds good but it’s not). Many recognize there’s no substitute for ‘learning on the job’ — but <strong>how do you get that data science job in the first place</strong>?</p><p>Instead, I recommend building up a public <strong>portfolio of simple, but interesting projects</strong>. You will learn everything you need in the process, perhaps even using all the resources above. However, you will be highly motivated to do so and will retain most of that knowledge, instead of passively glossing over complex formulas and forgetting everything in a month. If getting a job as a data scientist is a priority, this portfolio will open many doors, and if your topic, findings or product are interesting to a broader audience, you’ll have more incoming recruiting calls than you can handle.</p><p><em>Here are the steps I recommend. They are optimized for maximizing your learning and chances to get a data job.</em></p><p><strong>1. Pick a topic you’re passionate or curious about</strong>.</p><p>Cats, fitness, startups, politics, bees, education, human rights, heirloom tomatoes, labor markets. Research what datasets are available out there, or datasets you could create or obtain with minimal effort and expense. Perhaps you already work at a company that has unique data, or perhaps you can volunteer at a nonprofit that does. The goal is to answer interesting questions or build something cool in a week (it will take longer, but this will steer you towards something manageable).<br>Did you find enough to start digging in? Are you excited about the questions you could ask and curious about the answers? Could you combine this data with other datasets to produce original insights that others have not explored yet? Census data, zip-code or state level demographic data, weather and climate are popular choices. Are you giddy about getting started? If your answer is ‘meh’ or this feels like a chore already, start over with a different topic.</p><p><strong>2. Write the tweet first.</strong></p><p>(<em>A 21st century, probabilistic take on the scientific method, inspired by Amazon’s “write the press release first” practice and, more broadly, the Lean Startup philosophy</em>)</p><p>You’ll probably never actually tweet this, and you probably think tweets are a frivolous avenue to disseminate scientific findings. But it’s essential that you write 1–2 sentences about your (hypothetical) findings *before* you start. Be realistic (especially about being able to do this in a week) and optimistic (about actually having any findings, or them being interesting). Think of a likely scenario; it won’t be accurate (you can make things up at this point), but you’ll know if this is even worth pursuing.<br>Here are a few examples, with a conversational hook thrown in:</p><ul><li><em>“I used LinkedIn data to find out what makes </em><a href="https://blog.linkedin.com/2011/09/01/entrepreneur-data"><em>entrepreneurs</em></a><em> different — it turns out they’re older than you think, and they tend to major in physics but not in nursing or theology. I guess it’s hard to get VC funding to start your own religion.”</em></li><li><em>“I used Jawbone data to see how weather </em><a href="https://jawbone.com/blog/how-temperature-affects-exercise/"><em>affects</em></a><em> activity levels — it turns out people in NY are less sensitive to weather variations than Californians. Do you think New Yorkers are tougher or just work out indoors?”</em></li><li><em>“I combined BBC obituary data with Wikipedia entries to see if </em><a href="https://medium.com/@jasoncrease/was-2016-especially-dangerous-for-celebrities-79d79b9fae02#.et5utm1e3"><em>2016</em></a><em> was as bad as we thought for celebrities.”</em></li></ul><p>If your goal is to learn particular technologies or get a job, add them in.</p><ul><li>From <a href="https://blog.insightdatascience.com/fellow-spotlight-shelby-sturgis-ad24757982de#.lcqhie33q">Shelby Sturgis</a>: “<em>I built a web application to help teachers and administrators improve the quality of student education by providing analytics on school rank, progress on test scores over time, and performance in different subject areas. I used MySQL, Python, Javascript, </em><a href="http://www.highcharts.com/"><em>Highcharts.js</em></a><em>, and </em><a href="http://d3js.org/"><em>D3.js</em></a><em> to store, analyze, and visualize </em><a href="http://star.cde.ca.gov/"><em>California STAR</em></a><em> testing data.</em>”</li><li>“<em>I’ve used TensorFlow to automatically colorize and restore black and white photos. Made this giant collage for Grandma — best Christmas ever!</em>”</li></ul><p>Imagine yourself repeating this over and over at meetups and job interviews. Imagine this in USA Today or story or Wall Street Journal (without the exact technologies; a vague “algorithm” or “AI” will do). Are you boring yourself and having trouble explaining it, or do you feel proud and smart? If the answer is “meh”, repeat step 2 (and possibly 1) until you have 2–3 compelling ideas. Get feedback from others — does this sound interesting? Would you interview somebody who built this for a data job?</p><p>Remember, at this point you have not written any code or done any of the data work yet, beyond researching datasets and superficially understanding which technologies and tools are in demand and what they do, broadly speaking. <em>It’s much easier to iterate at this stage</em>. It sounds obvious, but people are eager to jump into a random tutorial or class to feel productive and soon sink months into a project that is going nowhere.</p><p><strong>3. Do the work.</strong></p><p>Explore the data. Clean it. Graph it. Repeat. Look at the top 10 most frequent values for each column. Study the outliers. Check the distributions. Group similar values if it’s too fragmented. Look for correlations and missing data. Try various clustering and classification algorithms. Debug. Learn why they worked or didn’t on your data. Build data pipelines on AWS if your data is big. Try various NLP libraries on your unstructured text data. Yes, you might learn Spark, numpy, pandas, nltk, matrix factorization and TensorFlow — not to check a box next to a laundry list, but because you *need* it to accomplish something you care about. Be a detective. Come up with new questions and unexpected directions. See if things make sense. Did you find a giant issue with how the data was collected? What if you bring in another data set? Ride the data wave. This should feel exciting and fun, with the occasional roadblock. Get help and feedback online, from Kaggle, from mentors if you have access to them, or from a buddy doing the same thing. If this does not feel like fun, go back to step 1. If the thought of that makes you hate life, reconsider being a data scientist: this is as fun as it gets, and you won’t be able to sustain the hard work and the 80% drudgery of a real data job if you don’t find this part energizing.)</p><p><strong>4. Communicate</strong><br>Write up your findings in simple language, with clean, compelling visualizations that are easy to grasp in seconds. You’ll learn several data viz tools in the process, which I highly recommend (it’s an underrated investment in your skills). Have a clean, interesting demo or video if you built a prototype. Technical details and code should be a link away. Send it around and get feedback. This being public will hold yourself to a higher standard and will result in good quality code, writing and visualizations.</p><p>Now, do it all again. Congratulations, you’ve learned a lot about the latest technologies and you now have a portfolio of compelling projects. Send a link to the hiring manager on your dream data science team. When you get the job, send me a Sterling Truffle Bar.</p><p>(Note: This article originally appeared on <a href="https://www.quora.com/session/Monica-Rogati/1">Quora</a> and <a href="http://www.forbes.com/sites/quora/2017/01/20/whats-the-best-path-to-becoming-a-data-scientist/#419f45716905">Forbes</a>)</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F7b0181%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F7b0181%2F&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/748a06e9019942f2677bba917ef55804/href">https://medium.com/media/748a06e9019942f2677bba917ef55804/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f8074232608e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/good-audience/how-do-i-become-a-data-scientist-f8074232608e">How do I become a data scientist?</a> was originally published in <a href="https://medium.com/good-audience">Good Audience</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A happy little binary tree: learning to code from Bob Ross]]></title>
            <link>https://medium.com/read-think/https-medium-com-mrogati-bobross-2bc000f0d4bd?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/2bc000f0d4bd</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[teaching]]></category>
            <category><![CDATA[computer-science]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Sat, 29 Oct 2016 15:42:00 GMT</pubDate>
            <atom:updated>2016-11-11T16:30:51.222Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/236/1*43tgo7Hskb7vOksksqejFA.jpeg" /><figcaption>Image credit: Julian Bergheim, AFAIK</figcaption></figure><blockquote>“If you have to do something over again, it doesn’t mean you’re bad, it just means you’re normal.” — Bob Ross</blockquote><p>Bob Ross has been famous for decades for his soothing videos that teach painting, one friendly little tree at a time. What made them special was watching the creative process as it happened, with mistakes calmly welcomed as ‘happy little accidents’ and turned into teaching opportunities. He gave viewers the confidence that they, too, can paint elaborate landscapes. This is why <strong>the best coding classes are like Bob Ross videos: making something beautiful in real time, with errors and detours —</strong> gently and gracefully corrected.</p><p>I had learned how to code in 6th grade, making stick figure animations in BASIC. In 9th grade, I was coding backtracking algorithms in Pascal (they involved cannibals and missionaries crossing a river. Don’t ask.) I had paid my C++ dues in CS101. But my understanding of computer science and coding fundamentally changed when I took my first Scheme class. (Scheme was the Scala of the ‘90s). Yes, functional programming can blow your mind, but I was also lucky to have the Bob Ross of computer science as my teacher. His name is <a href="http://www.bcl.hamilton.ie/~barak/">Barak Pearlmutter</a>, and he’s also a neural networks expert since the last time they were cool. His teaching style was unlike any I had seen before.</p><p>This was a time when everybody was lecturing and copying prepared code snippets onto transparencies or the whiteboard. Instead of Bob Ross videos, the lessons would look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*Vz46c7zaV6rQ1ldbYNTkpw.jpeg" /><figcaption>Image credit: <a href="http://knowyourmeme.com/memes/how-to-draw-an-owl">http://knowyourmeme.com/memes/how-to-draw-an-owl</a></figcaption></figure><p>Barak was different. He would bring his projector and laptop, sit down in front of the class, open up his Emacs window, and start coding from scratch. He usually had a task in mind, but, like his artsy counterpart, would be flexible about the details.</p><p>Unlike in linear, unrealistic step-by-step tutorials, you’d get to have a front seat to the authentic coding process. You’d get to watch somebody who was extremely good, but surprisingly imperfect. You’d see the typos and logic errors being made, discovered and fixed; you’d see how he would test bits and pieces of code separately, isolating any problematic behavior to its simplest incarnation. <strong>You could see the code evolving from one line to a fleshed-out program — not like a progress bar, slowly and linearly, but like an embryo, sprouting new organs and limbs and resorbing others.</strong></p><p>For students new to coding, seeing the non-linear evolution of a piece of code was crucial. It was demystifying the process, unbundling it, exposing the thought behind every step, normalizing bugs. <strong>The teacher’s imperfection was key to instilling confidence, especially in those not as naturally confident.</strong> Today, there are many well-made, interactive demos and tutorials, but they still don’t make enough mistakes. The editing process is throwing away the best part: the messy “rest of the owl”.</p><p>If you missed your chance at taking a Bob Ross-style CS class — it’s probably not too late. You might want to try pair-programming with a more experienced engineer for a similar effect. This <a href="http://ask.metafilter.com/239161/Happy-little-binary-trees-Is-there-a-Bob-Ross-of-coding">thread</a> also has some great pointers. And if you’re teaching others how to code, embrace your mistakes: they’re just happy little accidents.</p><h3>Think other people would benefit from this article? Click the heart to help them find it ❤</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2bc000f0d4bd" width="1" height="1" alt=""><hr><p><a href="https://medium.com/read-think/https-medium-com-mrogati-bobross-2bc000f0d4bd">A happy little binary tree: learning to code from Bob Ross</a> was originally published in <a href="https://medium.com/read-think">ThinkGrowth.org</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Literacy, Product Design and the Many-Faced God]]></title>
            <link>https://medium.com/art-marketing/data-literacy-product-design-and-the-many-faced-god-cf8339e035a?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/cf8339e035a</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[google]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[apple]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Tue, 18 Oct 2016 21:53:37 GMT</pubDate>
            <atom:updated>2016-10-18T22:22:53.554Z</atom:updated>
            <content:encoded><![CDATA[<p>It’s official — AI has gone mainstream. For data scientists, this good news, although there’s some ambivalence about its <a href="https://medium.com/@mrogati/machine-learning-ai-and-the-emperors-vest-32b7bdd99b58#.2nx75lb8r">hype</a> and <a href="http://www.npr.org/2016/09/12/493654950/weapons-of-math-destruction-outlines-dangers-of-relying-on-data-analytics">ethical applications</a>. Companies are taking note — and too often, they hire a data scientist to sprinkle magic machine learning dust and bring the company into the AI era. That, of course, doesn’t work — unless the company is ready to <a href="https://twitter.com/mrogati/status/788106887170363392">embrace data literacy</a>.</p><p>Case in point: Apple recently <a href="https://twitter.com/rsalakhu/status/788041946325479424">hired Ruslan Salakhutdinov</a>, a prominent deep learning expert and professor at Carnegie Mellon, my alma mater, to build an AI center of excellence. This is exciting; it’s heartening to see Apple finally paying attention to AI and machine learning — but is it too little, too late?</p><p>Building a <a href="https://jobs.apple.com/us/search?pr=1#mixes&amp;ss=52662972&amp;t=0&amp;so=&amp;lo=0*USA&amp;pN=0&amp;openJobId=52662972">team</a> that’s doing ‘<em>cutting-edge research in deep learning, machine intelligence, and artificial intelligence’</em> is not easy — not in this hiring environment. But <strong>infusing data thinking throughout a company is orders of magnitude harder</strong>. This matters, because data thinking permeates your products and can make them feel “smart” — or not.</p><p>Here’s an example.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*F4O5K-iuNTN9iySt2PUz5Q.jpeg" /><figcaption>I have many faces.</figcaption></figure><p>MacOS Sierra came with a new version of Photos — one with better face recognition and better organization, not unlike Google Photos (which I am a big fan of). I’ve enthusiastically started using it, as it was slowly (and I do mean slowly — over a month) indexing my pictures. Face recognition was disappointing — I was identified as 100 different persons, and another member of my family as over 200. I manually — and painstakingly — merged them, only for more alter-egos to appear as indexing continued. Pretty bad user experience, overall — and one I didn’t have with Google Photos, where I maxed out at a dozen alter-egos. (Sunglasses happen.)</p><p>Fortunately, this is the part that Ruslan and his team of machine learning PhDs can fix with better face recognition algorithms that can match Google’s. Unfortunately, that’s the part that’s easy compared to an even bigger problem: <strong>data as a second class citizen is reflected in product design</strong>. (This is a widespread issue. I’m picking on Apple because I just clicked the wrong thing and lost my selection of 77 faces I was going to merge.)</p><p>Consider this: Apple Photos detects hundreds of faces, and I diligently merge &amp; label them. Now, Photos knows a) how many times a person appears in my photos; b) what % of my photos include this person (for every year of my life!); c) how many times I appeared in a photo with them; d) how many times one of my ‘favorites’ appears in a photo with them. You can even imagine building a co-occurence graph and iteratively calculating the strength of those edges, then visualize it à la LinkedIn <a href="https://blog.linkedin.com/2011/01/24/linkedin-inmaps">InMaps</a>. (Imagine, but <a href="https://xkcd.com/356/">don’t actually do it</a>. )</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NtMM-iI9Ufi7awyWOv6dgw.png" /><figcaption>Now imagine 300 people instead of 5.</figcaption></figure><p>Bottom line — <strong>Apple Photos should know how important these people are.</strong></p><p>And yet — how does Apple Photos order these faces?</p><p><strong>Alphabetically.</strong></p><p>OK. Maybe they ran user studies and this was the least confusing and less cognitively taxing method. Maybe it’s the easiest way to scan for a person (search is not a first class citizen, either). Maybe it was a very deliberate decision to give more control to the user and explicitly select “favorites”. Maybe the product manager veto-ed an engineer who came up with a measure of importance because it sounded too convoluted. Maybe nobody thought of using these signals because it’s not part of the product/dev culture, or maybe they have — but decided to not even bother bringing it up. As a result, Photos does not feel “smart”. Walt Mossberg said the same thing <a href="http://www.theverge.com/2016/10/12/13251068/walt-mossberg-apple-siri-is-dumb">about Siri</a>.</p><p>But you don’t even need AI or sophisticated algorithms to make products feel ‘smart’ — not initially. Often, including in this case, simple counting and division is enough — if you think about it creatively. To do that, <strong>you need data thinking to be part of the culture and top of mind, not an after-thought</strong>.</p><p><em>PS: How does Google Photos solve this problem? It orders people by frequency, and lets you search for their names. No need to overthink it — and it’s just as easy to understand as alphabetical order.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cf8339e035a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/art-marketing/data-literacy-product-design-and-the-many-faced-god-cf8339e035a">Data Literacy, Product Design and the Many-Faced God</a> was originally published in <a href="https://medium.com/art-marketing">ART + marketing</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Rise of the Data Natives]]></title>
            <link>https://medium.com/@mrogati/the-rise-of-the-data-natives-8d1510fd856c?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/8d1510fd856c</guid>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Tue, 17 May 2016 17:14:42 GMT</pubDate>
            <atom:updated>2016-10-18T22:15:34.177Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A longer version of this article appeared in </em><a href="http://www.recode.net/2014/4/10/11625490/the-rise-of-the-data-natives"><em>Recode</em></a><em> two years ago, before Amazon Echo, Tesla Autopilot, Google Photos and the AI </em><a href="https://medium.com/@mrogati/machine-learning-ai-and-the-emperors-vest-32b7bdd99b58#.rolb235ec"><em>hype</em></a><em>. Since then, it has inspired a </em><a href="http://2015.datanatives.io/"><em>conference</em></a><em> that is now a yearly event, and we’re all becoming data natives (or at least really well-assimilated data immigrants).</em></p><p>A few years ago, YouTube was abuzz with <a href="https://www.youtube.com/watch?v=aXV-yaFmQNk">viral videos of toddlers</a> pinching magazines with their fingers as they would an iPad. These children were heralded as members of a new generation of <a href="http://edition.cnn.com/2012/12/04/business/digital-native-prensky/"><em>digital natives</em></a>: People who grew up surrounded by computers, shaped by always-on technology and the Internet.</p><p><strong>We are now witnessing a new revolution — that of <em>data natives</em> who expect their world to be “smart” and seamlessly adapt to their taste and habits.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LgSZoekF4_xQvZ02WdWv-g.png" /></figure><p>While digital natives ask what they can do <em>with</em> technology, data natives are more concerned about what technology can do for <em>them</em>. It’s not enough for a magazine to be digital and responsive; it should be personalized and context-sensitive.</p><ul><li>Digital natives program their thermostat. Data natives expect the thermostat to program itself.</li><li>Digital natives use the Starbucks mobile app. Data natives want the app to <a href="http://blog.adamnash.com/2013/08/27/my-letter-to-starbucks-mobile/">know their favorite drinks</a> — and when to suggest a new one.</li><li>Digital natives use a cloud-connected baby monitor. Data natives expect their baby monitor to automatically calculate <a href="http://www.theatlantic.com/magazine/archive/2012/05/the-data-driven-parent/308935/">crying percentiles</a> based on millions of other babies.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/708/1*05iNwlKaUOey8yGftlx_lg.png" /><figcaption>Autocorrect miss on a tweet featuring “the SCIENCE guy”, “US Chief Data SCIENTIST”, and “learning some sconce”, a phrase with zero Google hits (vs. 12K for “learning some science”). We’re not there yet.</figcaption></figure><p>There are frustrating times ahead for data natives. In a constantly connected, data-rich world, all of our expectations are evolving quickly. Autocomplete and autocorrect are everywhere — and we <a href="http://www.damnyouautocorrect.com/">make fun</a> of them when they don’t work. We’re frustrated when our GPS doesn’t autocomplete, or when it shows us a restaurant 1,000 miles away — so we’ve replaced it with Google Maps or Waze. Now, we wonder why it hasn’t already learned our preferred route.</p><p>We “settle” for self-parking, lane-following cars with adaptive cruise control when what we really want are self-driving ones. Your “smart” watch wakes you up at 3 am to tell you its battery is running low. Your Roomba is gathering dust — and not in a good way. We sigh and repeat “operator” when the cheery phone bot is telling us to state our problem again. True, these are prime examples of first-world problems, but it’s fair to ask: <em>If we haven’t yet fixed the small things, how can we be trusted with the innovations that would really enhance all our lives?</em></p><h4>The good news is that data natives’ frustration with not-so-smart technology is exactly what will make the promise of artificial intelligence a reality.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/852/1*vNBBWD1QU7_SYoQmSgfqHQ.png" /></figure><p>Technical skills are crucial, but they are not enough to satisfy the data natives’ expectations. The tide is now turning toward making data and sophisticated algorithms invisible. Top-notch design and user interaction, combined with data and algorithms, are what make products feel smart. These two sides of the data-product coin are reinforcing each other, creating a virtual cycle. Easy-to-use devices or software that seamlessly integrates into your life generates higher volume, better-quality data, while personalized experiences enabled by data are big contributors to a product’s ease of use.</p><h4>A good user interface offsets algorithm shortcomings, and helps technology make the leap from novelty to indispensable.</h4><p>Humans, after all, are highly adaptable if they can reap the benefits. Yes, this means you have to speak punctuation marks when you dictate your text messages, but while doing so, you’re training the system to infer them in the next version. While data natives expect technology to be smart, they understand that a little human input goes a long way.</p><p>Just like the children playing “iPad” on their magazines, too many of us still experience a parent-to-child relationship with technology — you need to tell it what to do very specifically, and correct often.</p><p>Data natives, on the other hand, are working toward a more grown-up relationship with technology — a technology that anticipates your needs while requiring little input beyond passive observation. This is the holy grail of artificial intelligence. We’ll get there, albeit after suffering through a few more years of Clippy 2.0, DeepClippy and ClippyBot.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8d1510fd856c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Pitfalls — the Startup Edition]]></title>
            <link>https://medium.com/@mrogati/data-pitfalls-the-startup-edition-fe1eebcc39da?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/fe1eebcc39da</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Wed, 11 May 2016 15:58:43 GMT</pubDate>
            <atom:updated>2017-08-06T17:12:20.681Z</atom:updated>
            <content:encoded><![CDATA[<h3>Data Pitfalls —the Startup Edition</h3><p><em>(A version of this article originally appeared in the </em><a href="http://www.amazon.com/Lean-Analytics-Better-Startup-Faster/dp/1449335675"><em>Lean Analytics</em></a><em> book, which I recommend based on your click history.)</em></p><p>Startups love data. Over the past years, they’ve become better and better at collecting it at scale, analyzing it and building <a href="http://firstround.com/review/this-is-how-you-build-products-for-the-new-generation-of-data-natives/">data products</a>. As data and its role grows and evolves, a few common pitfalls emerge:</p><p><strong>1. Assuming the data is clean: </strong>Cleaning the data you capture is often most of the work, and the simple act of cleaning it up can often reveal important patterns. Is an instrumentation bug causing 30% of your numbers to be null? Do you really have that many users in the 90210 zip code, or born in 1900? Check your data at the door to be sure it’s valid and useful.</p><p><strong>2. Not normalizing:</strong> Is Chicago the most popular wedding destination? Technically, yes, if you simply count the number of people flying in for a wedding. But if your goal is to compile a list of “popular wedding destinations”, you need to consider the <em>total </em>number of people flying in — otherwise, your list will consist of <a href="https://www.tripit.com/blog/2011/08/destination-wedding-week.html">airline hubs &amp; big cities</a>.</p><p><strong>3.</strong> <strong>Excluding outliers: </strong>Those 21 people using your product more than a thousand times a day? Either they’re your biggest fans, or bots crawling your site for content. Whichever they are, ignoring them would be a mistake. (50 years ago pulsars were <a href="https://twitter.com/NASAblueshift/status/894180788736020480">discovered</a> by not ignoring noise in the data.)</p><p><strong>4. Including outliers:</strong> While those 21 people using your product a thousand times a day are interesting from a <em>qualitative</em> perspective, because they can show you things you didn’t expect, they can be problematic when building models. You probably want to exclude them when building data products; otherwise, the ‘…you may also like…” feature on your site will have the same items everywhere. At LinkedIn, we called this the “<a href="http://mitultiwari.net/docs/papers/browsemap.pdf">Obama effect</a>”.</p><p><strong>5.</strong> <strong>Ignoring seasonality: </strong>“Whoa, is ‘Intern’ the fastest growing job this year? Oh, wait, it’s June.” Failure to consider time of day, day of week, and monthly changes when looking at patterns leads to questionable decision-making.</p><p><strong>6.</strong> <strong>Ignoring size when reporting growth:</strong> (An investor favorite!) Context is critical. When you’ve just started, technically, your dad signing up just doubled your user base.</p><p><strong>7.</strong> <strong>Data vomit:</strong> A dashboard isn’t much use if you don’t know where to look and <a href="http://www.kaushik.net/avinash/digital-dashboards-strategic-tactical-best-practices-tips-examples/">what to do next</a>.</p><p><strong>8. Metrics that cry wolf:</strong> (A DevOps favorite!) You want to be responsive, so you set up alerts to let you know when something is awry in order to fix it quickly. But if your thresholds are too sensitive, they get “whiny”’ — and you’ll start to ignore them.</p><p><strong>9.</strong> <strong>The “Not Collected Here” syndrome: </strong>Mashing up your data with data from other sources can lead to valuable insights. Do your best customers come from zip codes with a high concentration of sushi restaurants? This might give you a few great ideas about what experiments to run next — or even influence your growth strategy. At Jawbone, we looked at how <a href="https://jawbone.com/blog/how-temperature-affects-exercise/">weather affects activity</a> levels, and applied those lessons to making the product better.</p><p><strong>10.</strong> <strong>Focusing on noise:</strong> We’re hardwired (and then programmed) to see patterns where there are none. It helps to set aside the vanity metrics, step back and look at the bigger picture. After all, if you’re building something where there was nothing before, you’re making it <em>infinitely </em>better.</p><p>TL;DR: When it comes to data, it’s important to ask the right questions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/184/1*eJg08G4WmuKUxxzLL3w--A.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fe1eebcc39da" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine Learning, AI, and the Emperor’s Vest]]></title>
            <link>https://medium.com/chatbot-news-daily/machine-learning-ai-and-the-emperors-vest-32b7bdd99b58?source=rss-b31aba7f4828------2</link>
            <guid isPermaLink="false">https://medium.com/p/32b7bdd99b58</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Monica Rogati]]></dc:creator>
            <pubDate>Wed, 27 Apr 2016 16:38:05 GMT</pubDate>
            <atom:updated>2016-11-03T15:08:32.225Z</atom:updated>
            <content:encoded><![CDATA[<p>Machine Learning, AI, and the Emperor’s Vest</p><p>Those of us who work in data science and artificial intelligence have a love/hate relationship with hype. We’re excited by self-driving cars, machines understanding complex images, and computers beating humans at Go (if not StarCraft). On the other hand, we’ve heard stories of the last ‘AI winter’ and we fear that hype (and the inevitable trough of disillusionment that follows) is setting us up for another one. We know <strong>machine learning is math, not magic</strong>, and we don’t want to be left holding the bag when someone declares that the AI emperor has no clothes.</p><p>As always, the truth is more nuanced. <strong>It’s not that the AI emperor has no clothes; he’s just wearing a vest. </strong>A practical, warm, rugged vest full of proven technology with concrete applications. This is a good thing — just like AI, a vest is a lot more useful than the bejeweled imperial garb we pretend to see when talking about building artificial brains.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/484/1*Fzx_hMVELHxhndQliTQ-Uw.png" /><figcaption>Image credit: scottevest.com</figcaption></figure><p>Machine learning researchers and practitioners are justifiably nervous: the mainstream press now talks about “deep learning” mimicking the human brain, AI-powered bots stealing everybody’s jobs, and AI being a “new life form that’s a potential threat to our survival as a species”. But experts in the field have declared the brain metaphor tenuous at best, and preventing AI from turning evil akin to preventing overpopulation on Mars. They are legitimately concerned with using technology ethically and responsibly, but also resent being cast as the inept, wild-haired scientists unleashing evil into the world (or, even worse, accused of selling AI snake oil).</p><p><strong>Machine learning is already changing humanity.</strong> Much of that change is good — from better spelling correction to personalized medical diagnosis. Some of it is bad: machine learning can unwittingly perpetuate racism, hate and discrimination, mirroring the worst parts of humanity. But we need to stop thinking about AI as magic. <strong>What we call machine learning is a collection of mathematical techniques </strong>— albeit advanced enough that they’re starting to be indistinguishable from magic. This practical AI vest may not be as exciting as a magical suit. But it serves us well, and while we’re working on making the cargo pants we don’t have to worry about being sweet summer children caught in the next AI winter with only imaginary clothes on.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=32b7bdd99b58" width="1" height="1" alt=""><hr><p><a href="https://medium.com/chatbot-news-daily/machine-learning-ai-and-the-emperors-vest-32b7bdd99b58">Machine Learning, AI, and the Emperor’s Vest</a> was originally published in <a href="https://medium.com/chatbot-news-daily">Chatbot News Daily</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>