<?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 Scribble Data on Medium]]></title>
        <description><![CDATA[Stories by Scribble Data on Medium]]></description>
        <link>https://medium.com/@scribbledata?source=rss-c79d8fa01ccf------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*75SMvpxD8-OQTQpy.jpg</url>
            <title>Stories by Scribble Data on Medium</title>
            <link>https://medium.com/@scribbledata?source=rss-c79d8fa01ccf------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 10:58:01 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@scribbledata/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[Everything You Wanted To Know About Machine Learning (But Were Afraid To Ask)]]></title>
            <link>https://medium.com/@scribbledata/everything-you-wanted-to-know-about-machine-learning-but-were-afraid-to-ask-1b758aa1faf5?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b758aa1faf5</guid>
            <category><![CDATA[explanation]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Fri, 07 Dec 2018 10:54:25 GMT</pubDate>
            <atom:updated>2018-12-10T05:52:45.524Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RuiyLX5EU0eY41GEQW7EuA.jpeg" /><figcaption>Gerri Van der Walt on Unsplash</figcaption></figure><p>We’ve all been there — a meeting that starts off innocuously enough but suddenly takes a hard left towards a blindspot, and to our horror, we’re in a tiny minority, being assailed by terms we’ve never heard but having to nod along cheerily to. Well, if that blindspot is Machine Learning, this post will prep you well for that next board meeting, or water cooler conversation.</p><p><strong><em>1. What is Machine Learning?</em></strong></p><p>The conversations used to be about Data Analytics and Data Science, so what’s this ML business about? Nothing spooky, <strong>ML is just an umbrella term for throwing computing power and statistics at data with the intention of predicting or labeling stuff.</strong> And when you hear about Artificial Intelligence, well, know that ML is a subset of AI. The difference between data analytics and machine learning is primarily the sophistication of the methods used and the volume of data. Both are applications of statistics in different forms but machine learning helps answer persistent questions with limited human involvement (after the machine learning model has been built and fine-tuned for the purpose), while data analytics is much more high touch, requiring people to help answer questions.</p><p>In short, companies have data, say about how their customers behave (buying, complaining, etc). They build a machine learning model to study that data, or to be ‘trained’ on that data, and then they send the model off into the real world so that when the next set of data about a customer’s behaviour comes in, the model can sort that behaviour into a particular pattern that it knows to recognize, or make predictions as to what will happen next with the customer.</p><p><strong><em>2. What are types of machine learning models?</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pscLDrw72QQazBUOSHPjsA.jpeg" /><figcaption>Kiven Zhao — Unsplash</figcaption></figure><p><strong>a. Supervised Learning</strong> — Remember how we said some models had to be trained on data? Well, supervised learning is when the input data to a model has been already labeled with what the right answer is. If we show a model a picture of a cat, and we also tell the model it is a cat, the model begins to form an idea of the FEATURES* of a cat (its catness, in purely technical terms). It helps to also train it with photos of non-cats that are identified as such. The end result is that when the model is rolled out into the real world, and a picture of a cat is shown to it, the ML model will be able to identify the photo as having a cat in it.</p><p><strong>b. Unsupervised Learning</strong> — This is when the input data has no labels on it. Here the model’s job is to look at the entirety of the data coming in and identify patterns in the data if any exist. If the algorithm chosen for the model is sound, you may get the data to be bucketed into neat clusters that can be further reasoned about. If an unsupervised learning model is shown a number of different pictures of animals, while it may not be able to tell you which of them is a cat, it will sort the pictures of cats and zebras into different buckets to tell you that these are different animals.</p><p><strong>c. Semi-supervised Learning</strong> — Takes bits of both types above. Here, typically the end goal is to make predictions or come to conclusions (like in supervised learning), but the data coming in is a mix of labeled data and unlabeled data. The model here needs to look at the data to figure out underlying patterns so that it can organize the data into buckets/clusters and then make a prediction.Again, with the cats. The SSL model sees a stream of photos of animals, some correctly pre-labeled as being cats, and sorts the photos into cats, and zebras, and hippos, and tells you which of the animals are cats.</p><p>As in our example above, about customer behaviour data, if we wanted to classify these customers into frugal and heavy spenders, an unsupervised learning method might create multiple buckets of such customers by various features. If we added in some labeled data, that classified these customers into the two buckets we were interested in, our semi-supervised algorithm would effectively be tweaked to provide the binary type of clustering (frugal and heavy spenders) that we were interested in.</p><p>Still with us? Here comes some name dropping that should be helpful.</p><p><strong><em>3. Tools and frameworks whose name and functions you should be familiar with</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z1LcyvpA0sQGsazS9skeeQ.jpeg" /><figcaption>Matt Artz on Unsplash</figcaption></figure><p>The actual process of deploying ML requires certain tools. This section lists a few of the most popular ones and the role they play in the ML ecosystem</p><p><strong>Python</strong></p><p>The programming language of choice for most ML folk. Guess what? It gets its name from Monty Python’s Flying Circus!</p><p><strong>Pandas</strong></p><p>It’s cute, it’s cuddly, and also a software library that extends the Python language, used for data modeling, or to get the input data lined up nicely.</p><p><strong>Jupyter notebooks</strong></p><p>This is a web app that developers use to code in — to clean data and build the scratch versions of their models to test them on their own machines, before sending them into the real world</p><p><strong>General ML Frameworks like NumPy, Scikit-Learn</strong></p><p>NumPy is an extension of Python that lets developers do Dr. Strange levels of manipulation on large, unwieldy data</p><p>Scikit-Learn sits atop NumPy and has the actual ML algorithms that are used in building the models.</p><p><strong>Neural Network ML Frameworks: TensorFlow, PyTorch, Keras, Caffe2</strong></p><p>These are libraries to help deploy ML at scale, acting as the go-between for the models and the powerful hardware that are used to run these models.</p><p><strong><em>4. How Organizations are using ML</em></strong></p><p>Businesses today use ML to solve a large spectrum of problems like figuring out which customers to target with which offers; predicting weather, and using that to estimate sales; detecting fraud; looking at scans and medical records to determine health issues that may elude a doctor, and on and on.</p><p>Irrespective of the industry, organizations are identifying persistent problems or opportunities, putting together teams of data scientists and engineers to build ML to tackle these, and then taking action on the predictions generated.</p><p>Is there something else in the ML domain you’ve come across or heard about that a google search isn’t throwing light on? Comment here, and a fellow scribbler will get back to you on it</p><p>— — — — — — — — — — — — — -</p><p><em>*</em><strong><em>Scribble Data is a FEATURE ENGINEERING company</em></strong><em>, helping businesses take their Machine Learning models from lab to the real world. Check us out at </em><a href="http://www.scribbledata.io/"><em>scribbledata.io</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b758aa1faf5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Should Data Scientists Be Excited Or Worried About The New Privacy Laws?]]></title>
            <link>https://medium.com/@scribbledata/should-data-scientists-be-excited-or-worried-about-the-new-privacy-laws-a0b35070c770?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/a0b35070c770</guid>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[srikrishna-committee]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[gdpr]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Mon, 01 Oct 2018 06:21:27 GMT</pubDate>
            <atom:updated>2018-10-01T06:21:27.916Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xZqNpAjJI5t6zgWqBWPQtw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/photos/fPxOowbR6ls?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Matthew Henry</a> on <a href="https://unsplash.com/@matthewhenry?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>The General Data Protection Regulation (GDPR) legislated and passed by the European Union has sent ripples around the world, and depending on who you ask, this could either spell apocalypse, the workings of a nanny state, or a very positive step towards consumer privacy. The direct objective of such a ruling is to give control to individuals over their personal data and to simplify the regulatory environment for international business by unifying the regulation within the EU. The larger context over here is essentially a philosophical one — people have their right to privacy, as in the real world, so too in the digital realm. Closer home in India, while our IT laws lag a bit behind, the Supreme Court did make a landmark judgment in 2017 which unambiguously stated that privacy is a fundamental right to the citizens of India, and the Srikrishna Committee has tabled their draft of a data protection bill, that <a href="https://www.business-standard.com/article/economy-policy/here-are-the-hits-and-misses-of-india-s-draft-data-protection-bill-118092600142_1.html">still needs refinement</a>. Crucially, this will certainly have bearing on matters related to personal data and data privacy when the time comes. Now that we have the legalese out of the way, let’s explore what this means and entails for data scientists.</p><p><strong>The Fundamental Conflict</strong></p><p>Of late, the paths of data scientists and privacy advocates have diverged — there is a schism in methodology. Data Science has the goal of acquiring new data and finding new uses for existing data. Privacy advocates strive to minimize data collection while data scientists strive to maximize it. Privacy advocates are moving to decrease unexpected uses of data, while data scientists are incentivized to increase them. What has come to light in the past few years are the ways in which data scientists are able to cull insights and learnings from what was thought to be innocuous data. Projects have been carried out which link sensitive but anonymized data to specific individuals, reveal the gender and/or ethnicity of individuals based on social media likes, retrieve personal records of individuals based on camera footage on the street, fingerprint cell phones based on cell tower check-ins, etc. To comply with GDPR, both data scientists and privacy advocates need to work together to walk the tightrope of the rights to one’s privacy whilst ensuring there are no economic setbacks once data is more regulated.</p><p><strong>What This Means For Data Scientists</strong></p><p>The new laws will certainly require a whole new learning curve for data scientists in the following processes:</p><ul><li><strong>Ability to collect data</strong>: Individuals will need to give express consent for what data are collected and will need to be informed as to why the data are being collected</li><li><strong>Ability To Use Data</strong>: It will become necessary to get express consent for each application of personal data</li><li><strong>Ability To Transfer Data To And From Third Parties</strong>: Stiff regulatory fines will certainly produce an environment where corporations are very reluctant to buy, sell or share data that may be personal. In addition, the right to privacy/erasure regulation may have strong implications on data sharing. In short, we can expect a drying up of certain data sources.</li></ul><p><strong>Why We Need To Look At This Positively</strong></p><p>It’s hard not to get carried away by the enormous potential of big data, but, as so many data leaks have taught us, there are people that are the victims of these leaks. Real human beings, with rights to their own data and privacy. Often, newly-minted data scientists have played fast and loose with privacy and consent — think culling data from social media (relevant here: <a href="https://www.cbsnews.com/news/facebook-said-to-use-peoples-phone-numbers-for-ad-targeting/">Facebook monetizes your 2FA phone numbers</a>); but with the new regulations percolating through, data science teams will have to work with narrower data sets, but those that come with a higher degree of consent.</p><p>Data Science will have fixed boundaries around it again. While this might read to some as a more restricted playing field, it will channel the focus back on process, sound technique, tools, and relevance to business context. This move bodes well for the art of data science and authentic practitioners of the art. Their work will come to be appreciated more and we can expect recommendations, targeted offers and the like coming to them either with better focus and pertinence. Most importantly, questions around the ethics of data scientists, that could otherwise distract from the results they produce, will start to recede.</p><p>As it stands, a more promising, rich and ethical digital landscape is on the horizon.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a0b35070c770" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why your business doesn’t have to wait, to start giving back]]></title>
            <link>https://medium.com/@scribbledata/why-your-business-doesnt-have-to-wait-to-start-giving-back-376d9d1c6620?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/376d9d1c6620</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[philanthropy]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[company-culture]]></category>
            <category><![CDATA[startup-lessons]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Wed, 05 Sep 2018 09:42:01 GMT</pubDate>
            <atom:updated>2018-09-05T09:42:01.122Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i0_5lD4caUdepVZkhx0lkQ.jpeg" /><figcaption>Abhishek Pawar — Unsplash</figcaption></figure><p><em>“If you’re in the luckiest 1% of humanity, you owe it to the rest of humanity to think about the other 99%.” </em>— Warren Buffett</p><p>W.B. has given away more than he has left. In fact, he has pledged to give 99% of his wealth. It gives us pause. When talk of CSR and philanthropy are brought up in most offices, there’s a general lull in the air — the process seems like an imposition for most who just consider it another thing to tick off a sundry list. But the act of giving is powerful. It’s not just about the feel good. It’s about empowerment, about creating networks, and about working towards higher purposes, all of which act as fuel for your business’ own success. Let’s dive in.</p><p><strong>Giving Is Enlightened Self Interest</strong></p><p>Large sections of societies do not hold the capital that companies often do. We must realise that markets cannot exist in social vacuums, and giving is a way to create bridges for aspiration gaps — the gaps that exist between those that have ability, and the opportunities that then tap into this ability. Giving is also a societal safety net for when misfortune strikes. In some sense, giving is the most practical way of keeping what’s left.</p><p><strong>Giving Makes For A More Diverse Work Space</strong></p><p>We’d all love a diverse work space. Homogenous work environments fail to bring out creative solutions and better problem solving. Most companies have a pretty predictable composition of people — managers hire people with whom they perceive to have shared backgrounds and mindsets, despite wanting to be diverse. By making a concerted effort to give, companies stand to learn about worlds that are distanced from their experiences and develop a better and more empathetic view of the eco-systems that they operate in. The kind of employees that value this mindset, and are drawn to companies that proactively give, add new colour to the overall employee palette.</p><p><strong>Network Effects</strong></p><p>Besides being fiscally prudent (most Governments will provide tax concessions for philanthropy), there ultimately is nothing more satisfying than giving and seeing your goodwill translate into impact. It’s equally important that smaller companies like ours, despite their limited resources, start practising this right at the outset. It could be a small amount, or perhaps it’s a donation of time or kind, but it still counts.</p><p>To use a cliché, what goes around, comes around. Acts of giving provide purpose to employees and businesses alike. They make the work one does more meaningful than simply profits or pay cheques. This engenders enthusiastic conversations outside the workplace. In turn, it sends signals in the marketplace, both to customers and potential employees alike, about the kind of business and the employee peer group they can expect when they work with you.</p><p>One way to encourage this mindset among employees, rather than simply as a top-down directive, is to provide matching programs — where the company matches (or multiplies) the amount gathered by employees — towards specific causes. At Scribble, we’re always looking for people that are well-rounded, and that are wondering how they can increase the radius of their sphere of impact.</p><p>Let’s talk. Say hello to us at Scribble</p><p>hello@scribbledata.io</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=376d9d1c6620" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to get the most out of your organization’s data: The mindset]]></title>
            <link>https://medium.com/@scribbledata/how-to-get-the-most-out-of-your-organizations-data-the-mindset-e5418881d185?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/e5418881d185</guid>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Mon, 03 Sep 2018 06:15:34 GMT</pubDate>
            <atom:updated>2018-09-03T06:15:34.582Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tw33Cam0j6AZBrAoNbb5PQ.jpeg" /><figcaption>Photo credits: Ricardo Angel Gomez</figcaption></figure><p>Every business is a data business And while this aphorism has been around for some time, what does this actually mean to enterprise stakeholders? What should key decision makers be valuing and excited about as they start to invest in analytics tools and ML/AI?</p><p>Here’s what we think are the most important aspects to embrace when it comes to data and enterprise.</p><p><strong>Appreciation for Opportunity</strong></p><p>Yes, it merits restating. Every business is a data business. No matter what industry your enterprise exists within, it has collected diverse sets of raw data over time. The opportunity exists on many levels:. What do you want to gain insights on? How is your data segmented? How many possibilities are there to find relevant data insights that match your business objectives? How many new goals and ideas are you willing to invest effort behind? Are your resources in place to set new (but possibly high ROI) goals and see them through? The bottom line: data is transformative, the ‘how’ of transforming is limitless, therefore the more focused you are on specific goals, the more data can work for you.</p><p><strong>Accepting the Inherent Challenges of Data</strong></p><p>Identifying game-changing insights from data mining isn’t a magic trick. Its very nature is iterative and exploratory. This appreciation of the labour, diligence and time spent on the fundamentals of getting the data right will not immediately offer ROI, but it’s critical to the maturity of an organisation. The process takes stock of how you’ve done business and what the results have been across many platforms, customer touchpoints, and metrics that are important for your business . This stock-taking is invaluable to the cultural foundation of your enterprise. It allows you to see where you’ve excelled and where you need more nuts and bolts, or elbow grease in the system. It also allows enterprises to see where more risks can be taken and which ones should be taken first according to your business objectives and capabilities. As enterprise stakeholders, an appreciation for the nature of data and the challenges it throws at you is needed to reap long-term benefits.</p><p><strong>The Right Set Of Skills</strong></p><p>Hiring and retaining talent is of paramount importance for long-term success. Agility and flexibility in tooling (both, the tools, and the mechanic’s skillset) of all kinds is absolutely critical. Retaining and growing data talent must be centered on finding people who are not only technically sound but well aware of best practices in development, operations and open source. The most productive teams build response systems at every level of data mining. Having an understanding of your data from various sources, at all levels of ingestion allows for better challenge management, as new bumps in the road emerge, and more diverse insights.</p><p>Let’s talk data. Say hello to us at Scribble</p><p>hello@scribbledata.io</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e5418881d185" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[3 Key Learnings from India’s Biggest Data Conference]]></title>
            <link>https://medium.com/@scribbledata/3-key-learnings-from-indias-biggest-data-conference-56dd7cc5ad33?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/56dd7cc5ad33</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Mon, 13 Aug 2018 06:59:17 GMT</pubDate>
            <atom:updated>2018-08-13T06:59:17.948Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bJAfxV4qB-R6ajA8GdSKQA.jpeg" /><figcaption>© denys-nevozhai-unsplash</figcaption></figure><p>July saw us at the HasGeek-organised<strong> Fifth Elephant conference</strong> that hosted diverse talks with data practitioners from Flipkart, InMobi, Swiggy and more. Our co-founder at Scribble Data, Venkata Pingali, helped HasGeek with the development of the program for the 2-day conference. As you might know already, Fifth Elephant events are community-powered and focus on developing innovation in the data and AI space in India.</p><p>We were particularly excited to learn and discuss the specific challenges that data-driven organisations were thinking about and how they were tackling them. We weren’t disappointed. Here are 3 key takeaways from the conference, that are strong signals in where the bleeding edge of applied data science (i.e. in industry, and in production) is at.</p><p>— — — —</p><p><strong>(1) Buzzwords Are Being Called Out For Being Buzzwords</strong></p><p>In the last couple of years, there’s been a certain way the industry has engaged with the wider audience — A cycle of buzzwords that were semi-understood but still coveted by key technology decision makers. We’d hear about resume-enhancing projects that implemented complex statistical modelling but couldn’t make it to production. The discussions at the conference let go of the pixie dust and grounded everyone in real-world data fundamentals. It was a refreshing way to take stock of the directions in which data engineering and ML are tackling challenges at various enterprise levels. Our takeaway is that the rubber is beginning to meet the road. Data science was always difficult and laborious. It is good to see a more realistic picture painted.</p><p><strong>(2) The vital role of Data Engineering in productionizing Data Science</strong></p><p>Speakers from organizations such as Samsung and Slack lent nuance to the boundaries of data engineering and data science. The two are interconnected and feed off each other, but understanding their individual roles in the larger game is vital to productionize successfully. Data science is flexible when it comes to experiments, but these don’t always see the light of day as live deployments. A salient challenge is with the data engineering that lays the pipelines (data integrity, reliability, frequency, et al).</p><p>The opportunities and challenges that Data Science and ML can solve are abundant, but for these solutions to go the distance, it is essential to build teams that value well thought out, scalable and robust systems and processes.</p><p><strong>(3) Unexpected Debuts and Sparkling New Opportunities</strong></p><p>There was a noticeable presence of new, somewhat unexpected entrants, such as Cummins (engines), Mahindra (electric vehicles and mobility), and Star TV (entertainment). The ‘spread’ of the companies getting into the data business is growing. Everyone seems to have mandates to grow their teams and infrastructure. Bangalore, for sure, retains its position as the hub for data professionals. Most of these new entrants are considering setting up offices here too, making Bangalore the Data Brain of the country.</p><p>— — — —</p><p>Companies today are hungry for data experts. Bangalore will be a top destination in hiring this category of skills, but what what will it take for the data world to ensure that analytics/data science/ML initiatives continue to gain ground rather than being thought of as experiments? We think it’s going to be about the people, and how they collaborate.</p><p>Specifically, this means <strong>more dialogue, and a willingness to share information </strong>— particularly when it comes to challenges and failures we’ve conquered or renegotiated. To this end, we particularly enjoyed the Fifth Elephant conference, for the platform it gave the speakers, including the breakout ‘BOF’ (birds-of-a-feather) sessions, and the general bonhomie between attendees during networking sessions. Here’s to more.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=56dd7cc5ad33" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Scribble Partners With Cloudera]]></title>
            <link>https://medium.com/@scribbledata/scribble-partners-with-cloudera-8c99ecacc58e?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/8c99ecacc58e</guid>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Wed, 17 May 2017 09:53:19 GMT</pubDate>
            <atom:updated>2017-05-17T10:30:15.875Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="http://www.scribbledata.io/">Scribble Data</a> is pleased to announce our technical partnership with <a href="http://www.cloudera.com/">Cloudera</a>.</p><p>Cloudera, Inc. (NYSE: CLDR) is the provider of a leading modern platform for machine learning and advanced analytics that is built on the latest open source technologies.</p><p>Scribble’s rapid data querying platform complements Cloudera’s data platform, and enhances customers’ ability to speed up preparation, discovery and advanced analytics of complex data.</p><p>“We are excited to partner with Cloudera. Scribble’s own tooling has its origins in the hadoop world. Our partnership aims to meet the growing needs of advanced and adhoc analytics, where data democratization will give businesses a competitive edge, if complemented with the appropriate tooling,” said Dr. Venkata Pingali, CEO of Scribble Data.</p><p>“We found specific opportunities in the data querying space where it made sense to integrate with the Cloudera platform. The objective of the partnership is consistent with Scribble’s own play, which is to enable businesses to ask and answer much richer questions of their data, and get back reliable answers quickly,” explained Indrayudh Ghoshal, who co-founded Scribble Data and heads business operations.</p><p>Further news about integrated offerings from Scribble and Cloudera will be made available in the near future. For any additional information, please reach us at <a href="/service/mailto:contact@scribbledata.io">contact@scribbledata.io</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/559/1*nIOAZXyQCV0BNXUcmJJCCw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8c99ecacc58e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Implications of WannaCry Ransomware on Data Ecosystem]]></title>
            <link>https://medium.com/@scribbledata/implications-of-wannacry-ransomware-on-data-ecosystem-9e7c67dfc501?source=rss-c79d8fa01ccf------2</link>
            <guid isPermaLink="false">https://medium.com/p/9e7c67dfc501</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[ransomware]]></category>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[security]]></category>
            <dc:creator><![CDATA[Scribble Data]]></dc:creator>
            <pubDate>Tue, 16 May 2017 09:55:01 GMT</pubDate>
            <atom:updated>2017-05-17T08:30:31.084Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/744/1*FvT2J9JtgOOOzw_BHiubSA.jpeg" /><figcaption>Blue Coat Photos (CC BY-SA 2.0) Source: <a href="https://www.flickr.com/photos/111692634@N04/18495846450/in/photostream/">https://www.flickr.com/photos/111692634@N04/18495846450/in/photostream/</a></figcaption></figure><p><a href="https://en.wikipedia.org/wiki/WannaCry_ransomware_attack">WannaCry</a> is a remarkable attack. It is the first large scale demonstration, large by number of machines, of the disruptive effect of an attack on data. It was predictable and straightforward. Data as a viable attack target has been established, and bitcoin as a way to monetize attacks through its global currency and marketplace. We can only expect more attacks on data going forward. The attacks might go beyond the ransom and be more adversarial.</p><p>The cost of data access will increase due to increased tooling, process, and other overheads that will be introduced as a result of this attack. Data scientists and tool providers should expect greater scrutiny on access, integrity and communication of data. The tool providers will face tougher questions on new vulnerabilities introduced by their tools. CDOs will institute organization-wide data discipline to reduce exposure.</p><h3>Why The Attack Was Successful</h3><p>The main threat, as articulated by the WannaCry creators themselves, was the loss of information. Many of the impacted organizations had deployed industry best practices, in addition to configuration and backup tools along with appropriate firewalls. Despite this, the reason the attack was still successful is because simple restoration as an approach did not provide the solution. This could be due to several reasons:</p><ol><li>Data could not be restored — There was too much data on each desktop that was valuable and not systematically tracked, such as emails, docs and powerpoint files.</li><li>The restoration process was not well established — Backup tools are complex, and geographical distribution or lack of automation made the cost of restoration prohibitive.</li><li>Data might have been copied — The bits may have left the building and therefore restoration did not eliminate the potential for data leaks</li></ol><p>There are two sources of vulnerabilities in modern data systems:</p><ol><li><strong>High system complexity and data management tool complexity</strong> — Organizations don’t invest resources to make the problem manageable. It is seen as a waste of time, effort, or money.</li><li><strong>Gap in approaches</strong> — The volume, use, and value of data is outpacing our ability to manage the downside. It is not even clear who will do it in organizations. The discussion about how to cope with vast amounts of data lying all over the enterprise has just begun.</li></ol><p>These persistent problems will require a longer discussion at the community level to find suitable approaches.</p><h3>Implications</h3><p>The short term implications will be, as always, driven by fear. The security spending will increase. It will make everybody’s life miserable by adding more complex software that few understand, and the story will repeat itself some time down the line.</p><p>We see the following responses to the attack from an analytics perspective:</p><p><strong>There will be Increased scrutiny of all data players</strong> including data scientists who access critical business data. Though the loss of internal organizational email, doc files etc result in loss of productivity, the bigger threat is that any loss of critical business data will result in legal and reputational challenges whose cost is much higher.</p><p><strong>Analytical process discipline will grow</strong>. While new techniques and technologies, including elements of AI and ML will continue to grow in the analytic space, the need to provide analytic discipline will grow. Data scientists will have to address questions regarding organization (where is the data residing?), integrity (is this data consistent?) and lineage (where did it come from?) beyond the usual questions on methodological soundness and value.</p><p><strong>Metadata will see growth</strong>. In order to even know the scope or impact of an attack, we first need an account of what is where (location) and what is it (content and context). Tools and processes will be introduced to track critical data throughout its lifecycle. The security cost of metadata is much lower than data itself, and it is usually generated by system and therefore amenable to automation. Both these will drive greater adoption of metadata in organizations.</p><p><a href="https://www.linkedin.com/pulse/data-shifts-power-within-organizations-venkata-pingali"><strong>CDO role will receive greater attention</strong></a>. Balancing the security concerns and business value of data, and instituting good organization-wide data discipline is a complex task and slightly different from the roles of CSO, CIO, and CEO.</p><h3>Conclusion</h3><p>The full scope and depth of the WannaCry attack will emerge over time. It has warned us about what is possible. This will slow down the democratization of data and analytical work unless we build novel tools and approaches to address the underlying fundamental drivers of risk including complexity and data discipline.</p><p><em>This post is authored by Venkata Pingali, CEO of </em><a href="http://www.scribbledata.io/"><em>Scribble Data</em></a><em>, a data automation company for analytics.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9e7c67dfc501" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>