<?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 Marcossponton on Medium]]></title>
        <description><![CDATA[Stories by Marcossponton on Medium]]></description>
        <link>https://medium.com/@marcossponton?source=rss-18fcb435c47e------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*NjT7u78fnJLNUOJL</url>
            <title>Stories by Marcossponton on Medium</title>
            <link>https://medium.com/@marcossponton?source=rss-18fcb435c47e------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 09:03:34 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@marcossponton/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[It’s not about being remote]]></title>
            <link>https://medium.com/@marcossponton/its-not-about-being-remote-fa53c18acd24?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/fa53c18acd24</guid>
            <category><![CDATA[emotional-intelligence]]></category>
            <category><![CDATA[remote-working]]></category>
            <category><![CDATA[emotions]]></category>
            <category><![CDATA[wellbeing]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Fri, 20 Mar 2020 17:18:46 GMT</pubDate>
            <atom:updated>2020-03-20T17:18:46.340Z</atom:updated>
            <content:encoded><![CDATA[<p><em>“If it makes you happy, so why the hell are you so sad”</em> / Sheryl Crow</p><p>After years of advocating for flexible schedules, this just doesn’t feel right. This global mandate to work remotely, as a child forced to play in a park, seems awkward and wicked.</p><p>How is productivity, that should have gone up, is becoming a growing concern? Why so many people are profusely sharing lists like “20 pro tips to work from home”. And yet, in the face of obvious advice such as ventilate your room, eat healthy snacks and dress as if you were going to the office, I’ve been staring at a blank spreadsheet for an hour, wearing pajamas in the dark, and thinking about going back to smoking.</p><p>What’s going on? Let’s face the elephant in the room. We would rather go to the office 10 hours a day than going through this quarantine.</p><p>We are emotional beings. Every piece of factual data we use to analyze situations and make decisions has already been filtered by our emotional system. Especially powerful facts, filtered through fear and anxiety, are carefully selected for our survival. And the rational listing of best practices is certainly not a good tool to facilitate a behavior change in a complex environment. Adding the complexity of adverse emotionality. Like a mother asking “You need to eat better” sounds almost contradictory if you are living in her basement, unemployed and getting over a nasty divorce. “Yeah Mom, I’ll eat my veggies as soon as I can put two pieces of my life back together”.</p><p>The point is, working from home is not even the point. We are not productive, and it feels awkward when our emotions are triggered to protect us with defensive reactions. When rather a rational behavior could make the difference.</p><p>We need to be more vocal in the value of expressing how we feel, at work or at home. We need to feel that we are part of a community, not only as a means to exercise and receive compassion, but also to transcend making sense of how reality impacts our minds and bodies.</p><p>The tools at hand, if you are not willing to engage in formal therapy, are still incipient but headed in the right direction. Anyway, even baby steps in identifying our emotions, what really hit us and how we are making sense of it, move us one step toward our goal. That is not working from home. It’s being happy with the work we do.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fa53c18acd24" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[An injured industry]]></title>
            <link>https://medium.com/@marcossponton/an-injured-industry-9b65f19dd07a?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/9b65f19dd07a</guid>
            <category><![CDATA[management]]></category>
            <category><![CDATA[work-engagement]]></category>
            <category><![CDATA[people]]></category>
            <category><![CDATA[tech-companies]]></category>
            <category><![CDATA[burnout]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Wed, 26 Feb 2020 19:00:17 GMT</pubDate>
            <atom:updated>2020-02-26T19:00:17.773Z</atom:updated>
            <content:encoded><![CDATA[<h4>happiness is no longer a nice to have</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_HhJncUWhZ8Sel4GYm0MCw.png" /></figure><p>Sports injuries are a fact of life. But if half the team on the field was injured, no one would expect them to perform at a professional level, and the coach would probably be looking for a new job very soon.</p><p>So why do we keep demanding top-level performance from our workplace teams when so many are injured with burnout or emotional strain?</p><p>We can criticize younger generations all day long, yet they come to us with quite valuable advice: being happy at work is a universal right everyone can claim. And it’s the right path to choose.</p><p>When not dealt with properly, the issues that stem from not seeing workers as humans run through the entire organization, taking a toll on everyone: workers, unable to afford the handicap in achieving what they want comparing to happy workers; and companies, trying to stay afloat, compared to those able to create an environment where collaborators can execute at their highest and actually enjoy what they do.</p><p>How do we get out of this mess? Technology is the only way to leverage emotional-development tools across the board, with universal access to all organizational levels, for companies of all sizes, in any region. Access to deep soft-skills development tools and actions is no longer a right reserved for those with leadership status. Workers and teams need to understand and deal with their emotions. Managers and higher ranks, as workers themselves, need to do the same and be able to work these tools into the relationships with subordinates and superiors.</p><p>This is not the sales pitch for yet another emotional-intelligence workshop. This is a wake-up call to acknowledge that intellectual work is more than intellectual tasks.</p><p>Everyone is responsible for making it work. This responsibility goes hand in hand with the power anyone has to change the status quo and make things better (insert Spiderman/uncle Ben quote here, although originally attributed to Voltaire). We need to team up across the board and commit our efforts as company leaders, investors, and influencers to this cause. From spreading the word to investing in products, or taking tangible action as direct leaders of technology teams.</p><p>We at <a href="https://yerbo.co/">Yerbo</a> are here just to help us all to tackle these problems. And I’m personally having my skin in the game.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9b65f19dd07a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Discusiones de equipos reales sobre Machine Learning en Ecommerce & Fintech]]></title>
            <link>https://medium.com/@marcossponton/discusiones-de-equipos-reales-sobre-machine-learning-en-ecommerce-fintech-91a72931cf05?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/91a72931cf05</guid>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[compliance]]></category>
            <category><![CDATA[canvas]]></category>
            <category><![CDATA[ecommerce]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Mon, 23 Jul 2018 13:04:31 GMT</pubDate>
            <atom:updated>2018-07-30T11:08:03.467Z</atom:updated>
            <content:encoded><![CDATA[<p>Los desafíos de comunicación entre equipos técnicos y otros más orientados a negocio no son extraños en el mundo del Fintech y/o del Ecommerce, sino todo lo contrario. Más allá del perfil de cada compañía, el ideal suele ser que todas las áreas trabajen en conjunto. La distancia se agranda siempre alrededor de tópicos de moda y con mucho ruido a su alrededor: Inteligencia Artificial y Machine Learning no son la excepción.</p><p>Hace un tiempo <a href="https://medium.com/@msponton/machine-learning-roi-c%C3%B3mo-modelar-el-valor-de-invertir-en-machine-learning-930741666a09">presentábamos</a> una herramienta que en <a href="https://www.machinalis.com/">Machinalis</a> utilizamos para articular la propuesta de valor de Machine Learning de modo que todos (sean técnicos o de negocios) se puedan sentar en la mesa a analizar y conversar colaborativamente.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ear19DbJiOa59n1jkyHmlg.jpeg" /></figure><p>En aquella presentación mencionamos que dicha herramienta puede usarse para storytelling . Para ilustrarlo, presentamos hoy el caso de una solución intensiva en Machine Learning, en los dos verticales mencionados al principio (Fintech &amp; Ecommerce). La solución, <a href="https://fictus.io/">Fictus</a>*, consiste en un servicio que, dada una imagen, predice la posibilidad de que esa imagen haya sido alterada o no.</p><h3>Una herramienta de Storytelling</h3><p>El resumen es sencillo. En el canvas propuesto se distinguen tres niveles. Los mismos sugieren tres formas de describir las claves del producto desde perspectivas diferentes. Hay que conectar, en un único relato, los componentes de cada nivel.</p><p>Veamos el ejemplo del primer nivel, leyendo de izquierda a derecha cada elemento. Un relato posible sería:</p><p>“<em>Para… [Objetivo de la Organización] creamos un producto que logre [Objetivo del Producto] en el cual por medio de Machine Learning podemos [Propuesta de Valor]. Lo utiliza [Stakeholder] para que el usuario final [Impacto en el usuario final]</em>”</p><p>Probemos con Fictus en concreto:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V7rd1sKUZth0NGvQwYEJ5g.jpeg" /></figure><p>El segundo nivel podría ser más breve aún.</p><p>“<em>Con [Data Input], hacemos [Propuesta de Valor] de modo que podemos generar [Data Output]</em>”</p><p>Probemos de nuevo con Fictus:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5NkhpOFCyEgW5Cx8PX6y8Q.jpeg" /></figure><p>El tercer nivel incluye todos los campos restantes y sigue más o menos la misma lógica. Por ser más extenso lo dejamos para que los lectores completen el relato por sí solos.</p><p>Esta es tan sólo una de las formas de utilizar el Canvas como una herramienta de storytelling. En nuestra experiencia, es la que nos ha resultado más útil.</p><p>¿Se pueden combinar u ordenar las piezas de otra manera en el relato?</p><p>Seguro.</p><p>Por ejemplo, en ocasiones ocurre que se quiere discutir haciendo especial foco en la UX. Entonces, empezamos por el impacto en el usuario final. Luego, si ahí no hay acuerdo, ¿para qué seguir discutiendo sobre la magia negra del Machine Learning?</p><h3>Mirando todo el Esquema y Decidiendo la Inversión</h3><p>Ahora bien ¿Importa más asegurarnos que no pasa *ninguna* foto que tenga alguna probabilidad de que haya sido alterada, por más mínima que sea? ¿Importa más que el sistema haga una moderación automática sólo de aquellos casos en los que esté muy convencido de que han habido modificaciones digitales? Ambas expectativas, en un contexto de recursos productivos escasos, implican definiciones que pueden marcar caminos muy distintos en términos de producto y en términos de inversión en Machine Learning. En otras palabras: es muy costoso y poco probable que podamos obtener una performance elevada persiguiendo ambos caminos a la vez, y es un momento típico en el que las discusiones pueden aparecer.</p><p>Imaginemos una compañía con un equipo de compliance responsable de la auditoría y revisión de la documentación que los usuarios tengan que subir como parte del proceso de signup. El equipo tiene métricas e indicaciones de productividad, por ejemplo, tiempo promedio que tienen que dedicarle a cada imagen. A este equipo poco le va a importar tener una predicción que cubra la mayor parte posible de los casos porque de igual manera van a revisar todo el conjunto de imágenes. Ahora bien, con certeza a este equipo le importa (y mucho) saber si hay ciertas imágenes que por alguna razón sobresalgan y a las que valga la pena dedicarles más tiempo y/o recursos de investigación. En el mismo caso, si los recursos de moderación son escasos y se sabe de antemano que no se van a revisar todas las imágenes, el mismo asistente puede “decidir” el orden de las imágenes que este equipo tiene que monitorear, empezando por aquellas que tengan más probabilidad de haber sido adulteradas y enfocando sus esfuerzos donde haya más probabilidad de fraude. Desde esta perspectiva, la inversión que se haga en el asistente basado en Machine Learning se inclina hacia mejorar tanto como sea posible la precisión de sus predicciones, y no tanto lo que en Machine Learning se conoce como Recall (que perseguiría aumentar tanto como sea posible la cantidad de casos del grupo total de imágenes que hayan sido alteradas, aún a riesgo de que aumenten considerablemente los casos de falsos positivos).</p><p>¿Toman estas decisiones sólo los equipos técnicos? ¿Las toman los de producto o negocio? En cualquier contexto, ¿sirve el ML Canvas propuesto para que ambos equipos puedan dialogar de manera más productiva con “el otro”? En nuestra experiencia sí, esperamos que en la de ustedes también ;-).</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2Ff%2F68bc97%3Fas_embed%3Dtrue&amp;dntp=1&amp;display_name=Upscribe&amp;url=https%3A%2F%2Fupscri.be%2F68bc97%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/d8021540cbf50434828deb5439e1cf04/href">https://medium.com/media/d8021540cbf50434828deb5439e1cf04/href</a></iframe><figure><a href="http://bit.ly/2BDa2xQ"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Axmr4bMScpjBU_WcOm263A.png" /></a></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/788/1*u6pz8xzjblHCXLMQUQGF3Q.gif" /></figure><figure><a href="https://twitter.com/planetaChatbot"><img alt="" src="https://cdn-images-1.medium.com/max/788/1*qQxzn3vV9Xu4lYMpuc5OOw.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=91a72931cf05" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine Learning ROI: How to model the investment value of Machine Learning?]]></title>
            <link>https://medium.com/@marcossponton/machine-learning-roi-how-to-model-the-investment-value-of-machine-learning-a29be1f2f827?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/a29be1f2f827</guid>
            <category><![CDATA[return-on-investment]]></category>
            <category><![CDATA[canvas]]></category>
            <category><![CDATA[innovation]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Tue, 12 Jun 2018 17:20:32 GMT</pubDate>
            <atom:updated>2018-06-22T17:15:44.255Z</atom:updated>
            <content:encoded><![CDATA[<h4>A dialogue and analysis tool for distinct teams</h4><p><em>Bizman tells something to which Techie seems useless. Techie responds with answers to which Bizman interprets as objections which would never make their business, their salaries, profit.</em></p><p>This mini parody and its variants repeat themselves continually when business orientated teams interact with technical teams. There are those who have the innate ability to build bridges and are central for everybody, even though they are not usually recognized for their value.</p><p>Nowadays, with the fashion and hype behind Artificial Intelligence and Machine Learning the point doesn’t lose validity, on the contrary: the SciFied promises of many marketing teams — irresponsible?- in every conference generate magical expectations that rarely end in something other than disenchantment…..and this is when everyone loses.</p><p>At <a href="https://machinalis.com/">Machinalis</a> we have worked for around a decade implementing Machine Learning solutions. And it has to be said: we have often been protagonists of this hype. Today, with a bit of a trail and scars, we are extra specially careful with this interaction — the moment when everyone can contribute the best of their talents and achieve effective science with real business value….and rules. For this reason, one of the tools we have developed and now use is the Machine Learning Value Proposition Model. We have found other tools (some <a href="http://www.louisdorard.com/machine-learning-canvas/">recently</a>) but other then encourage us to invest in the topic, we have not found them truly useful: fundamentally because of the impracticality in the early stages of the decision making process to invest in Machine Learning. These early stages are more sensitive and generate potential conflict between the teams mentioned at the beginning of the article, also in these stages we have found more value dedicating effort in generating shared communication tools.</p><p>The Canvas we present today is essentially a communication tool, interactive, which motivates the organization of questions and benefits for all sides. In the best case scenario it becomes a storytelling tool. Whether you are a business person who wants to seed a business idea and intuit that Machine Learning can add value, or your are on the technical side and want to establish common definitions useful for business people and the rest of the team, we hope that this (long) post helps: we share a short description of each field in the outline.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w0hXmVB5p9Suz2-fnuzOdA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*UwuHMHpGYuyble72cPsGNg.jpeg" /></figure><p><strong>Machine Learning Value Proposition</strong></p><p>The What and How. The task at hand. What problems do we resolve with Machine Learning? Typically with Machine Learning, predictions.</p><p>It is important to identify what specific contribution we are making before developing a story. It is not only identifying “the problem,” rather the factor or trigger that makes this proposal inevitable. A problem to predict the delivery date of a shipment can be explained from the point of view of precision and buffers, yet a trigger goes beyond this, towards generating positive expectations among users.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*enUERFoIzpvCkqL4JIlz-w.jpeg" /></figure><p><strong>Organizational Objectives</strong></p><p>The organization or company search behind this investment. The objective abstract depends on many factors: typically, the more concrete and coherent they are with the rest of the model elements, the better.</p><p>From a technical point of view, the organizational objectives pose a work horizon (scope), which permits us to trace from the proposition to what mobilizes the campaign, whether it is cost effectiveness, profit growth, or market share. From the story point of view, the objective poses the first point of empathy. We don’t begin a story for Doctors without Borders with “if you really want to make money…”. And it is important because of its functionality, a Machine Learning project can nourish different objectives without entering in conflict. The objectives also become contextualized if Machine Learning tackles a problem that is core for the company or not. Or in the worst case scenario, if it attacks a problem which comes into conflict with the objective. For example: If Machine Learning is used to detect fraud in account generation for a platform, and at the same time the platform’s objective is to grow its active user base, our story should not focus on the speed with which the objective is met, but rather on how the Machine Learning solution eliminates the risks of not achieving it in the proposed manner.</p><p>However, not being associated with the company objectives at some level does not make the project automatically non-viable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*3M3uTJfAffCc3I7z3K4l_w.jpeg" /></figure><p><strong>Product Objectives</strong></p><p>This is the space where the multiple leaderships and interests of the project battle it out. In terms of Machine Learning, this is the space where Metrics are defined, one of the principal resources of the technical teams. It is the point where the product make or break definitions are explicit. In our experience, a central point is that the objectives are spoken of technically (for example, precision/recall) but rather in a way all stakeholders understand without the need of a dictionary. Otherwise, the risk is run of having to waste a half hour of every meeting in reviewing the meanings of precision and recall.</p><p>The product objectives delineate the relative value of the Machine Learning contribution for the organization and the Machine Learning contribution to the product in question. For example: If a project plans to replace human classification of NSFW videos for one based on Machine Learning, we could generate an attractive reduction of costs for some stakeholders. At the same time, the classifier could have a level of precision that is not in line with the product objectives or introduce unacceptable risks, generating a conflict. A cohesive story sets up a thread between the organizational and product objectives and the Machine Learning value proposition.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*abOhNSfXtNmJoaktbDiGvw.jpeg" /></figure><p><strong>Stakeholders</strong></p><p>The actors directly involved in the use and consumption of the Machine Learning product (normally internal areas of your business). Concrete steps, behavior, stakeholder KPI’s that will be affected by the product. In what manner these steps are represented in the data which stakeholders use and generate daily. Specific interactions and types of stakeholders related to the system.</p><p>We should understand the impact of the Machine Learning solution on the journey of our stakeholder. It is not the same to implement a moderation solution of product commentaries in an online catalog for the community manager than the legal team, because what they have to win and lose varies substantially and with it the focus of the solution. In the first case, an omitted moderation could strengthen engagement (even with the cost of obliging the payment of a small fine over copyright for example). In the second, a series impact on the business objectives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*0iJXidoXce8DHURms_ujpg.jpeg" /></figure><p><strong>End User Impact</strong></p><p>Some Machine Learning solutions pass inadvertent to the final user. Others are the essence of interaction. Knowing how we impact the user is a point of departure of the whole proposal, and thus its location on the canvas. How a malfunction impacts a user is another key point. Even something apparently inoffensive, like a recommendation, can have a negative impact on the final interaction, and so, impact a business decision.</p><p>Distinct angles can signify more or less interaction by users and so corresponding viability. (What happens if we need each user to give feedback for 100 days straight for the predictive model to function? Is it a reasonable time from the UX perspective?)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*RXT0pvgZVfR61ehwPbC14A.jpeg" /></figure><p><strong>Data Input</strong></p><p>The resources you plan to use and process with Machine Learning. Possibly the list is not restricted to data, and could include heuristic previews, model previews, rules considered as data, etc. In a nutshell, if you don’t have a technical background it is one of the aspects, along with Product Objective and System Output, in which more effort should be put to generate precise definitions so as to make pitching the idea to the technical team viable (and so they give faster and more usable answers).</p><p>The success rate of a Machine Learning project is associated with the quality and availability of relevant data. This factor should be explicit when telling the story, or we fall into the magical belief of an algorithm that, with a bit of sketchy data can provide an extraordinary result. Has it happened to you to be in a conference where a sales team promises solutions too fantastic to be of this world? Has it happened that someone returned from a conference inspired by a sales team to ask for the development of a product as if you were Amazon…without the resources of Amazon? Here we should include suppositions of style “how do we know the user purchase history for the last year” or “how can we detect emotions based on security camera images”. Many stories break when they get to this point and turns out that the intrinsic data needed is not available or can not be generated. It is a good moment to make a cost evaluation of the data, who is their owner and what policies of governance pertain.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*XfeULvhcFB5K8EAU-KZ-Jg.jpeg" /></figure><p><strong>System Output</strong></p><p>It could be information, a concrete number, even the taking of a decision: the closer to the last the more complex is the product to build, but maybe there is more value to be gained. In either case, it is extremely useful to define the format in which the output will be generated and the mechanism and format to interact with other systems, so as to facilitate analysis and shrink reworking.</p><p>The data resulting from a Machine Learning solution is, in story terms, where we explain the gains leaving the evaluation up to the listener. Through the output data we finish validating our understanding of the problem, with the possibility that it may open in each case. In explaining that our solution can determine if an anonymous user is or not a millennial with tech brand loyalty with a couple of inputs, we let the receiver fill in blank spaces, not only in terms of the value proposition, but also in the impact potential in other business problems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*igjqn6AgH6bIuVFJU1mtWA.jpeg" /></figure><p><strong>Competitive Landscape</strong></p><p>A reference of the characteristics that the competition has in relation to the task at hand is one of the most relevant elements we have found for the moment of defining the value proposition. Whether it be in the search for parity in your product capacities or the contrary, for differentiation, the main benefit is in the power to exemplify the desired direction: enables more precise discussions and definitions of technical viability, aligns with the business strategy, anticipates friction with the users, etc.</p><p>The analysis of the competitive environment does not only consist in identifying other options to our solution, but also to know what those competitors have already installed. Some competitors may not even belong in the domain. Since Amazon and the later adoption by many ecommerce, the idea of free-returns is now the norm expected by buyers. If we consider penalizing, charging or not accepting returns we put ourselves in a situation of conflict of expectations born from the competition.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*uLjW8-mnJYIgF3H5HYDjhQ.jpeg" /></figure><p><strong>Structural Restrictions. Costs and Debts.</strong></p><p>The costs taken on by building this product: the current and future ones. A mature vision of the investment in systems or ML intensive products implies conscientious decisions about the debt we take on to validate a value proposition, not always “visible” at the beginning of the product unless energy is dedicated to identify them (e.g. “they lack a data culture”).</p><p>Estimating the cost of development of a solution is not only associated with coding time. Some solutions can be simple to develop and costly to test. The story to tell should reflect what technical concessions a MVP of the solution entails so as to gel correct expectations. If we believe that the point of inflection in the story is 10K transactions, we can base our model for doubts sake to support 20K, later needing to be recalculated (but with the principal hypothesis verified).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*uNXjPZxxuKmdZEfa3LZ6Ew.jpeg" /></figure><p><strong>Business Impact</strong></p><p>The impact on business is direct or indirect? The organizational and product objectives are not always transparent in relation to the benefit which the business generates. Also the same combinations of Organizational Objectives + Product Objectives + Value Proposition can pursue an impact or other in the business: for example, increase the value of the company vs increase the cashflow.</p><p>What will be the business impact of our solution? Is it saving or earning? Capitalize-able? Is it a vitamin, something to gain an opportunity or a painkiller, something to resolve a need? If we have a tangible impact ( a conversion increment aligned to the revenue objective, for example) we can support our story on the company objective. If the impact is more diffuse and looks to generate engagement, we can associate the story with the positive impact on interactions in the user journey.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*DEDcwWfYyUXXN2-1P3S6Bw.jpeg" /></figure><p><strong>Growth</strong></p><p>Moving away from stakeholders considerations and from the impact on specified user journeys, it is not strange that a Machine Learning intensive product has the potential to expand and grow toward other stakeholders and other corners of users journeys. To articulate in a unique tale the connection between the current value proposition, the expected impact on the business and the growth possibilities, helps that all actors involved can share the direction laid out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*QpxBOSfRdCuk8qxh8ZvbGQ.jpeg" /></figure><p><strong>Evaluation and Real Time Monitoring</strong></p><p>Legacy of previous models, the idea of this component in the Canvas is to make explicit what factors we observe to guarantee the health of the system. It is not uncommon to observe technical factors such as stuck models, data source corruption and unstable relationships between training vs production (in fact we have internal tools which I hope we can eventually share with the community), nevertheless in this point we intend a slightly different focus: share within the team, technical or not, which of the components mentioned above will be the factors to observe for the value proposition to keep being the coherent axis within the Canvas.</p><p>Implies a decision about the risk you wish to manage being online.</p><p>If you got here, Congratulations! It is the moment to try it out. Fill in at least 5 fields (with the Value Proposition as the only non-negotiable) and tell someone “from the other side” a story with those five fields. It wouldn’t surprise me if the others feel that you speak their language and the value analysis flows.</p><h4>More where this came from</h4><p>This story is published in <a href="http://blog.usejournal.com/">Noteworthy</a>, where thousands come every day to learn about the people &amp; ideas shaping the products we love.</p><p>Follow our publication to see more stories featured by the <a href="https://usejournal.com/?/utm_source=usejournal.com&amp;utm_medium=blog&amp;utm_campaign=guest_post">Journal</a> team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a29be1f2f827" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine Learning ROI: Cómo modelar el valor de invertir en Machine Learning?]]></title>
            <link>https://medium.com/@marcossponton/machine-learning-roi-c%C3%B3mo-modelar-el-valor-de-invertir-en-machine-learning-930741666a09?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/930741666a09</guid>
            <category><![CDATA[roi]]></category>
            <category><![CDATA[innovación]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[canvas]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Tue, 29 May 2018 14:19:43 GMT</pubDate>
            <atom:updated>2018-06-22T07:46:03.100Z</atom:updated>
            <content:encoded><![CDATA[<h4>Una herramienta de diálogo y análisis para los distintos equipos</h4><p><em>BizMan le dice al Techie algo que a Techie no le sirve de nada. Techie le plantea a Bizman respuestas que Bizman interpreta como objeciones que jamás harán que el negocio, que paga los sueldos de ambos, avance.</em></p><p>Esta mínima parodia y sus variantes se repite comúnmente cuando equipos más orientados a negocios interactúan con equipos más orientados a temas técnicos. Están los que tienen la habilidad natural de tender puentes y son de un valor central para todos, aún cuando no siempre sean reconocidos por ello. Hoy, con la moda y el ruido detrás de Inteligencia Artificial y Machine Learning el punto no pierde validez sino todo lo contrario: las promesas cercanas a ciencia ficción de muchos equipos -¿irresponsables?- de marketing en cada conferencia generan expectativas mágicas que raramente terminan en algo distinto al desencanto… y es cuando todos pierden.</p><p>En <a href="https://machinalis.com/">Machinalis</a> trabajamos hace alrededor de una década implementando soluciones de Machine Learning. Y hay que decirlo: hemos sido protagonistas muchas veces de este ruido. Hoy, con un poco de recorrido y cicatrices tenemos mucho y especial cuidado en esta interacción, momento en donde todos pueden aportar lo mejor de sus habilidades y lograrse ciencia con efecto y valor real para el negocio… siguiendo reglas de negocio. Para ello, una de las herramientas que en su momento desarrollamos y hoy utilizamos, es un <em>Modelado de Propuesta de Valor de Machine Learning</em> (Machine Learning Value Proposition Model). Esencialmente es una herramienta de comunicación, de interacción, que motiva a organizar las preguntas y aportes de todos lados . En el mejor de los casos, termina en una herramienta de <em>storytelling</em>. Sea que sos una persona de negocios que quiere plantear una idea de negocios e intuís que con Machine Learning se puede entregar más valor, o que sos una persona técnica y querés establecer definiciones comunes útiles para las personas de negocio y para todo tu equipo, esperamos que este (largo) post te ayude: compartimos una mínima descripción de cada campo del esquema.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yEAk_ko-RiPSTFSh0ukpBw.jpeg" /><figcaption>Machine Learning Value Proposition Model</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/359/1*3H7RUTf8bwBPWHU5nt4GKA.jpeg" /></figure><h3>Propuesta de Valor de Machine Learning</h3><p>El Qué y el Cómo. La Tarea a realizar. Qué problema resolvemos con Machine Learning? Típicamente en Machine Learning, una predicción.</p><p>Es importante identificar qué aporte específico estamos haciendo antes de desarrollar una historia. No es sólo identificar “el problema”, sino cuál es el <em>factor o trigger</em> que hace que esta propuesta sea inevitable. Un problema para predecir la fecha de entrega de un envío puede explicarse desde el punto de vista de la precisión y los buffers, pero un trigger va más allá, hacia poder generar expectativas positivas en nuestros usuarios.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*AYzznt3Zq1h7Gv9WD8inGA.jpeg" /></figure><h3>Objetivos de la Organización</h3><p>La búsqueda de la organización o compañía detrás de esta inversión. La abstracción del objetivo depende de muchos factores: típicamente, mientras más concreto y coherente con el resto de los elementos del modelo, mejor.</p><p>Desde un punto de vista técnico, los objetivos organizacionales plantean un horizonte de trabajo, nos permiten construir trazabilidad desde la propuesta hacia lo que moviliza a la compañía, sea rentabilidad, crecimiento en los ingresos o posicionamiento. Desde un punto de vista de la historia, el objetivo nos plantea el primer punto de empatía. No comenzamos una historia para Médicos sin Fronteras con “si realmente quieres hacer dinero…”. Y es importante porque desde su funcionalidad, un proyecto de Machine Learning puede alimentar diferentes objetivos sin entrar en conflicto. Los objetivos también contextualizan si Machine Learning ataca un problema core o no core para la empresa. O en el peor caso, si ataca un problema que entra en conflicto con un objetivo. Por ejemplo: Si se usa Machine Learning para detección de fraude en la generación de cuentas en una plataforma, y al mismo tiempo la plataforma tiene como objetivo hacer crecer su base de usuarios activos, nuestra historia debería enfocarse no en la velocidad con la que se alcanza el objetivo final, sino en cómo la solución de Machine Learning elimina riesgos de no lograrlo de la manera planteada.</p><p>Sin embargo, no estar asociado con los objetivos de la compañía a algún nivel no hace el proyecto inviable de manera automática.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*-__0utoYoc06Xcsp_3kDcg.jpeg" /></figure><h3>Objetivos del Producto</h3><p>Es el espacio donde los múltiples liderazgos e intereses presentes en el proyecto tienen una batalla. En términos de Machine Learning, es el espacio en donde se definen las Métricas, uno de los recursos centrales del equipo técnico. Es el punto en donde quedan explícitos los “<em>product make or break</em>”. En nuestra experiencia, un punto central es que los objetivos no pueden estar enunciados de manera técnica (por ejemplo, precision / recall) sino que expresarse de manera que todos los stakeholders entiendan sin recurrir cada vez a un diccionario. Caso contrario, se corre el riesgo de usar media hora de cada reunión de avance recordando los significados de <em>precision</em> y <em>recall</em>.</p><p>Los objetivos de producto dan idea del valor relativo para la organización del aporte de Machine Learning y del aporte de Machine Learning al producto en cuestión. Por ejemplo: Si un proyecto planea reemplazar una clasificación humana de videos NSFW por una basada en Machine Learning, podríamos generar una reducción de costos atractiva a algunos stakeholders. Al mismo tiempo, el clasificador podría tener niveles de precisión que no se alinean con los objetivos del producto o que introduzcan riesgos inaceptables, generando un conflicto. Una historia cohesiva plantea un hilo entre los objetivos organizacionales, los de producto y la propuesta de valor del Machine Learning.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*MzrEM4gpKPiJXH6ONmFU5Q.jpeg" /></figure><h3>Stakeholders</h3><p>Los actores directamente involucrados en el uso y consumo del producto intensivo en Machine Learning (normalmente, áreas internas de tu empresa). Pasos concretos, comportamientos, KPIs de los stakeholders que van a ser afectados por el producto. De qué manera esos pasos están representados en los datos que los stakeholders usan y generan diariamente. Interacciones específicas y tipos de stakeholders relacionados con el sistema.</p><p>Debemos entender el impacto de la solución de Machine Learning en el journey de nuestro stakeholder. No es lo mismo implementar una solución de moderación de comentarios de productos en un catálogo online para el gerente de comunidades de un e-commerce que para su responsable de legales, porque lo que tienen para ganar y perder varía sustancialmente y con ello el enfoque de la solución. En el primer caso, una moderación omitida puede aumentar engagement (aún a costo de obligarnos a pagar una pequeña multa por copyright por ejemplo). Para el segundo, impacta seriamente los objetivos de negocio.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*K5iTZ8_Z2cKsXmFURC1ALA.jpeg" /></figure><h3>Impacto en el Usuario Final</h3><p>Algunas soluciones de Machine Learning pasan inadvertidas al usuario final. Otras son la esencia de la interacción. Saber cómo impactamos al usuario es un punto de salida de toda la propuesta, y por eso su ubicación en el canvas. Cómo impacta en el usuario un malfuncionamiento es otro de los puntos claves. Hasta algo aparentemente inofensivo, como una recomendación, puede tener un impacto negativo en la interacción final, y por lo tanto, implica una decisión de negocio.</p><p>Distintos abordajes pueden significar más o menos interacción de estos usuarios y su correspondiente viabilidad (¿qué pasa si necesitamos que cada usuario dé feedback 100 días seguidos para que funcione el modelo predictivo? ¿es un tiempo razonable desde la perspectiva de UX?)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*p5UdyQQeOvq_fmZNBvQHZg.jpeg" /></figure><h3>Data Input</h3><p>Los recursos que pensás utilizar y procesar con Machine Learning. Quizás la lista no se restrinja sólo a data, y pueda incluir heurísticas previas, modelos previos, reglas consideradas como datos, etc. A los fines del relato, si no tenés formación técnica es uno de los aspectos, junto a Objetivo de Producto y al Output del Sistema, en donde más esfuerzo deberías poner, para generar definiciones precisas que hagan viable un pitching de la idea al equipo técnico (y que te den respuestas más rápidas y útiles).</p><p>Las probabilidades de éxito de un proyecto de Machine Learning están asociadas a la calidad y disponibilidad de datos relevantes. Este factor debe ser explícito al contar la historia, o caemos en el pensamiento mágico de un algoritmo que, con pocos datos de calidad dudosa puede proveer un resultado extraordinario. ¿Te sucedió alguna vez estar en una conferencia donde un equipo de ventas prometía soluciones demasiado extraordinarias para ser de este mundo? ¿Te sucedió que alguien llegó de una conferencia animado por un equipo de ventas a pedirte el desarrollo de un producto como si fueras del equipo de Amazon… pero sin los recursos de Amazon? Aquí debemos incluir supuestos del estilo “como sabemos el historial de compra del último año de este usuario” o “como podemos detectar emociones en base a imágenes de cámaras de seguridad”. Muchas historias se quiebran al llegar a este punto y comprobar que los datos que realmente se necesitan no están disponibles o no pueden generarse. Es un buen momento para hacer una evaluación del costo de los datos, quién es dueño de los mismos y cualquier política de data governance pertinente.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*zKQF0uxV71bOWu1crQvLxw.jpeg" /></figure><h3>Output del Sistema</h3><p>Puede ser desde información, un número concreto hasta la toma de una decisión: mientras más cerca de esto último, más complejo es el producto que hay que construir, pero quizás más valor haya para ganar. En cualquiera de los casos, es extremadamente útil definir el formato en que el output se va a generar y el formato y mecanismos para integrar con los otros sistemas, porque facilita el análisis y disminuye reproceso.</p><p>Los datos que resultan de la solución de Machine Learning son, en términos de la historia, donde explicamos la ganancia dejando la valuación a cargo del que la escucha. A través de los datos de salida terminamos de validar nuestro entendimiento del problema, con las posibilidades que esto abre en cada caso. Al explicar que nuestra solución puede determinar si un usuario anónimo es o no un millenial con lealtad a marcas de tecnología con un par de inputs, dejamos que el receptor complete los espacios en blanco, no sólo en términos de la proposición de valor, sino en el potencial para impactar en otros problemas del negocio.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*f9-hdSTKJU3HJMX0vdlWMw.jpeg" /></figure><h3>Competencia</h3><p>Una referencia a las características que tiene la competencia en relación a la tarea que hay que realizar es uno de los elementos más relevantes que hemos encontrado para el momento de definir la propuesta de valor. Sea por la búsqueda de paridad en capacidades de tu producto o todo lo contrario, de diferenciación, es un aporte clave el poder ejemplificador que tiene en cuanto a la dirección deseada: habilita discusiones y definiciones más precisas de viabilidad técnica, alinea con la estrategia de negocio, anticipa fricciones con los usuarios, etc.</p><p>El análisis del ambiente competitivo no consiste sólo en identificar otras opciones a nuestra solución, sino también reconocer, qué es lo que esos competidores ya han instalado. Algunos competidores pueden incluso no pertenecer al dominio. Desde que Amazon, y muchos ecommerce más tarde instalaron la idea de free-returns, esta es la norma esperada por los compradores. Si pensamos en penalizar, cobrar o negar retornos nos ponemos en una situación de conflicto de expectativas que nació en la competencia.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*7OaVX79SvLNh_d_4Q6pPFA.jpeg" /></figure><h3>Restricciones estructurales. Costos y Deudas.</h3><p>Los costos que se asumen construyendo este producto: los actuales y los futuros. Una visión madura de la inversión en sistemas o productos ML intensive implica decisiones conscientes sobre la <a href="https://papers.nips.cc/paper/5656-hidden-technical-debt-in-machine-learning-systems.pdf">deuda </a>que asumimos para validar la propuesta de valor, no siempre “visibles” al inicio del proyecto salvo que se dedique energía a identificarlas (e.g. “falta de una cultura de datos madura”).</p><p>Estimar el costo de desarrollar una solución no está asociada sólo a tiempo de codificación. Algunas soluciones pueden ser simples de desarrollar y costosas de testear. La historia a contar debe reflejar qué concesiones técnicas conlleva un MVP de la solución para fijar las expectativas correctas. Si creemos que el punto de inflexión en la historia está en 10K transacciones, podemos armar nuestro modelo en base a deuda para dar soporte a 20K, luego de lo cual sería necesario un refactoreo (pero con la hipótesis principal verificada).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*c-ip1Ci8KtYjYPStPATvZw.jpeg" /></figure><h3>Impacto en el Negocio</h3><p>El impacto en el negocio ¿es directo o indirecto? No siempre los objetivos organizacionales y de producto van a ser transparentes en relación al beneficio que genere al negocio, e inclusive las mismas combinaciones de Objetivos Organizacionales+Objetivos de Producto+Propuesta de Valor pueden perseguir un impacto u otro en el negocio: por ejemplo, aumentar el valor de la compañía vs. aumentar el cashflow.</p><p>¿Cuál será el impacto en el negocio de nuestra solución? ¿Es un ahorro o una ganancia? ¿Es monetizable? ¿Es una vitamina, algo para ganar una oportunidad o un analgésico, algo para resolver una necesidad? Si tenemos un impacto tangible (un incremento en la conversión alineado a un objetivo de revenue, por ejemplo) podemos apoyar nuestra historia en el objetivo de la compañía. Si el impacto es más difuso y busca generar engagement, podemos asociar la historia a su impacto positivo en la interacción con el usuario final. al user journey y cómo impactamos positivamente la interacción.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*BW-3gorq-fjhuoDi0Ymq1Q.jpeg" /></figure><h3>Crecimiento</h3><p>Alejándose de los stakeholders considerados al inicio y del impacto en el journey de los usuarios finales específicos, no es raro que un producto intensivo en Machine Learning tenga potencial de expansión y crecimiento hacia otros stakeholders y otros rincones del journey de los usuarios finales. Articular en un único relato la conexión entre la propuesta de valor actual, el impacto en el negocio esperado y las posibilidades de crecimiento, ayuda a que todos los actores involucrados puedan compartir la dirección trazada.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/258/1*2D0BwzzJZJzYed1YOfEn4Q.jpeg" /></figure><h3>Evaluación y Monitoreo en tiempo real</h3><p>Herencia de <a href="https://medium.com/louis-dorard/from-data-to-ai-with-the-machine-learning-canvas-part-i-d171b867b047">modelos anteriores</a>, la idea de este componente dentro del Canvas es explicitar qué factores vamos a observar para garantizar la salud del sistema. No es raro hoy observar factores técnicos tales como estancamiento de los modelos, corrupción de las fuentes de datos y relaciones estables entre entrenamiento vs. producción (de hecho tenemos herramientas internas que espero eventualmente podamos compartir con la comunidad), sin embargo en este punto intentamos un foco ligeramente diferente: compartir entre el equipo, técnico o no, cuáles de todos los componentes mencionados arriba van a ser los factores a observar para que la Propuesta de Valor sigue siendo un eje coherente dentro del Canvas.</p><p>Implica una decisión en relación al riesgo que se quiera gestionar estando online.</p><p>Si llegaron hasta acá ¡felicitaciones! Es momento de probarlo. Llenen al menos cinco campos (con la Propuesta de Valor como único no-negociable) y relaten a alguien “de la otra trinchera” una historia con esos cinco campos. No me sorprendería que los otros sientan que hablaron en su idioma y el análisis de valor fluya.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2Ff%2F68bc97%3Fas_embed%3Dtrue&amp;dntp=1&amp;display_name=Upscribe&amp;url=https%3A%2F%2Fupscri.be%2F68bc97%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/d8021540cbf50434828deb5439e1cf04/href">https://medium.com/media/d8021540cbf50434828deb5439e1cf04/href</a></iframe><figure><a href="http://bit.ly/2BDa2xQ"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Axmr4bMScpjBU_WcOm263A.png" /></a></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/788/1*u6pz8xzjblHCXLMQUQGF3Q.gif" /></figure><figure><a href="https://www.facebook.com/PlanetaChatbot/"><img alt="" src="https://cdn-images-1.medium.com/max/788/1*m2cnnbikTve7QdG-mXx57Q.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=930741666a09" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Science for Hire]]></title>
            <link>https://medium.com/@marcossponton/data-science-for-hire-64d0e44625e1?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/64d0e44625e1</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Wed, 04 Apr 2018 02:51:27 GMT</pubDate>
            <atom:updated>2018-04-04T02:51:27.526Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>6 tips on providing quality tailored data science services</strong></p><p>At Machinalis we work daily on projects that fall within the area known today as <em>Data Science</em>. Here are 6 tips and learned lessons for people who want to provide a sustainable data science service and don’t want to avoid the mistakes that we made.</p><ol><li>PROVIDE A SERVICE, NOT JUST DATA SCIENCE: Although it might sound obvious, if you love the technical aspects of your work, you risk falling in love with the challenge. The ultimate goal of a data science project is to build a tool that serves a purpose. All your efforts should be applied towards reaching this goal while keeping the client and the actual user of the product satisfied and you should avoid the temptation of exploring the subject further looking for interesting research avenues.</li><li>USE ADEQUATE EVALUATION CRITERIA: The evaluation criteria (from baseline to target performance) ought to be finite, completely understood and agreed up both by you and the client. All you do from then on is work towards optimizing the criteria. Any ambiguity in the criteria leads to re-work and loss of client confidence.</li><li>DESCRIBE “THE THING”: Find out if the object of analysis has certain desirable qualities (manageable size, ease of measurement, reasonable distribution, etc.) before starting any work. In this regard, spend some time trying to come up with a minimal descriptive statistic before tackling more complex approaches, always looking for unexpected results in the behavior you’re trying to predict. We all want to apply all the fancy new machine learning methods, but we have to be patient and make sure we’re that the universe we’re working on is even worth it.</li><li>GET THE DOMAIN SPECIALIST / PRODUCT OWNER INVOLVED: Validate your assumptions about the feasibility of the object of study with the product owner or domain specialist as soon as possible. Things that are strange to you might end up being normal. Did you find inconsistencies in the data? Share them with the team and the product owner. Need to reduce the size of the data? Validate the new data set with the domain specialist or product owner.</li><li>ILLUSTRATE / VISUALIZE: Visualization is essential. Minimalistic proposals that demonstrate the main idea are key. If you’re just starting in the area of visualization, rely upon existing products before investing on effort on your own solution.</li><li>SOMETIMES DATA SCIENCE IS JUST A STAGE: Sometimes in the context of a traditional software project, you might need to tackle a data science problem. Try to follow the tips that we mentioned, at least until you solve the specific problem. You can use whatever process works for you for the other non-data-science-related aspects of the project.</li></ol><p>To summarize: there’s nothing wrong with loving doing Data Science, and being passionate about it can get you a long way, but keep the objectives in focus and direct most of your effort towards it.</p><p>Thanks to Agustin Barto for his cooperation to write this article.</p><p><em>Originally published at </em><a href="https://www.machinalis.com/blog/data-science-for-hire/"><em>www.machinalis.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=64d0e44625e1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Don't buy Machine Learning]]></title>
            <link>https://medium.com/@marcossponton/dont-buy-machine-learning-da3305b63767?source=rss-18fcb435c47e------2</link>
            <guid isPermaLink="false">https://medium.com/p/da3305b63767</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[value-proposition]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Marcossponton]]></dc:creator>
            <pubDate>Tue, 03 Apr 2018 03:16:51 GMT</pubDate>
            <atom:updated>2018-04-03T03:18:43.711Z</atom:updated>
            <content:encoded><![CDATA[<p>During a conversation I had with Peter Norvig, we discussed about the kind of projects that we do at Machinalis and how strange does it feels to say that “we are a Machine Learning company”: In many projects, the amount of effort spent on R&amp;D on Machine Learning is usually a small fraction of the total effort, or it’s not even there because we plan it for a future phase after building the application first. At this moment, Norvig quoted a friend of him who said:</p><p><em>“Machine Learning development is like the raisins in a </em><a href="https://en.wikipedia.org/wiki/Raisin_bread"><em>raisin bread</em></a><em>: 1. You need the bread first 2. It’s just a few tiny raisins but without it you would just have plain bread.”</em></p><h3>So how do companies purchase this raisin bread?</h3><p>Typically, two things can happen:</p><p>A <strong>large company</strong>, at a higher level, decides to move away from standard tools, given that “our business and our data are different/peculiar”, incorporating Machine Learning or Data Science into their processes. In these situations they call us just to get some raisins.</p><p>A <strong>startup</strong>, after a few funding rounds, decides to transform the prototype they built to a production ready scalable MVP. In these situations they call us because they want the whole raisin bread, or perhaps some other kind of pudding which will need raisins at some point in the future.</p><p>Even if eating this is trendy, many people are not used to the new flavour, and it might not sit well with them.</p><h3>The trick is not buying “the Machine Learning thingy…”</h3><p>Unless you…</p><ul><li>… realize that by doing so you’re not getting into a typical software development project, but a *research* and development one. Even for some better-known problems that do not strictly follow this rule; the typical level of uncertainty is higher than in traditional development projects, and much higher if the expectations about the deliverable involve identify, design and/or achieve specific values for metrics like precision or accuracy.</li><li>… understand that there can be a large gap between the research paper that says “it’s possible” to the production ready software, sometimes so big that it can’t be covered today.</li><li>… view your predictive models created in the process as an asset. Your competitors don’t have access to this model, it is a piece of productive and potentially scalable knowledge which emerged from your very own combination of data and constraints. The team working on this is not wasting time as a “necessary evil”, but investing effort in the main asset of the process.</li><li>… do not expect that Machine Learning will be a black box that will produce magical results. A central principle of the process is “garbage in, garbage out” and if your goals are not clear, Machine Learning won’t make them any clearer.</li><li>… if you have a business background, you’re willing to incorporate some Machine Learning concepts that allows you to communicate a detailed vision of the problem you want to tackle from the business perspective to the team. The technical team will respond in kind by learning about the business side. You don’t want that, after having worked as separate silos, at the end the project you get a result that’s attractive from the scientific standpoint, but insignificant to your business (which means that you spent thousand and wasted time that your competitors have used better).</li><li>… consider that perhaps with 20% of the effort using standard and out-of-the-box algorithms you may get 80% of the expected results for the set metrics. Improving that result can be the remaining 80% of the effort… or even more.</li><li>… you’re aware that your Machine Learning team is usually pretty expensive, and you don’t want them having to deal with integrating their work with the existing production pipeline.</li><li>… your team is ready to start working with probabilistic results.</li><li>… do not buy it just because it’s popular: Startups in their early stages often do not need a full raisin bread, but just some bread prototype, perhaps a sample raisin and check if there’s interest on their street for a bakery.</li></ul><p>Do you think you can overcome all these obstacles? Congratulations, you have a good chance of enjoying a nice raisin bread in your business.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=da3305b63767" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>