译：Experfy模仿共享经济模型Uber 把自由数据科学家和分析项目的客户连接起来

36大数据专稿，原文作者：Alex Woodie 本文由36大数据翻译组-欧显东翻译，转载必须获得本站、原作者、译者的同意，拒绝任何不表明译者及来源的转载！

开始一个大数据分析项目是很艰难的，特别是对中型公司他们不能判断一个全职数据科学家的成本，不想处理传统的咨询公司。但一家名为Experfy发现成功通过模仿共享经济模型通过Uber推广，然后把自由数据科学家和分析项目的客户相匹配起来。

Experfy 在一年前出版了Harvard Innovation Lab目标是创建一个对高端咨询工作围绕数据分析的市场。 迄今为止， 该公司已经吸引了1200年数据科学家，他们通过客户创建方案和投标数据项目。

它的另一个迹象表明共享经济产生了数据的科学,， Experfy的CEO Harpreet Singh说“我们都是数据科学的Uber， 在美国如果你是一个中等规模的业务而且你想做数据分析，你可以雇佣一个人或者去咨询公司。但咨询公司非常昂贵，雇佣人工资会越来越高”。

到目前为止，Experfy的数据科学家已经成功完成大约60个项目，这些项目主要有从创建股票交易算法和构建推荐系统在Hadoop上，将Excel模型转换成软件即服务(SaaS)应用程序和创建仪表板在画面。

在某些情况下，Experfy的客户只是想要一个结果关于一个面向数据的问题，而在其他情况下，他们渴望的算法，是用R或Python，他们自己希望将实现。一些数据科学家为Experfy可能暂时充当项目经理工作，而团队的数据科学家可能参加一个客户的顾问委员会提供建议，对于审核技术方案由第三方供应商完成。

我们与很多与客户合作，Singh 告诉Datanami,“分析是一个复杂的事情。你不能只是去发布一个项目和期望神奇的发生。这里面有很多的辛勤工作。客户很欣赏这一点。他们想要的人将与他们合作从而帮助他们通过这个过程。”

Singh说“Experfy与客户一起对他们的项目进行精艺的描述在他们的项目发布到在线市场之前，一旦发布在线市场后，Experfy的数据科学家们可以提交自己的建议。只是阅读的建议是为这些客户解决问题，因为这些客户他们了解自己的问题，从不同的角度来看待。”

Experfy听取建议后，考虑客户的问题，帮助客户通过参与视频会议，一旦项目被选中时，中标Experfy双方一起工作，以确保工作按时完成,符合预期，然后佣金会由第三方支付，直到客户满意。Experfy收取20%的佣金，保证成功。

Experfy界面

一般没有项目会失败，为了增加成功的几率，Experfy鼓励客户搞较小的数据科学项目承诺。“更大的承诺,，意外着更大的项目，承受更大的风险”Experfy参与持续两到三周，平均花费10000美元至20000美元。然而这是没有限制的，最近的一个项目建立一个IBM Watson-powered应用程序花费了几个月的时间花费300000美元。

这一切就是要让数据科学进入一个新的企业。 “我有这个问题或需要做出这个决定或者我有一个业务问题,需要一个算法可以解决这个问题”。数据科学家可以很容易地来解决这个问题没有太多很多风险。“我们比麦肯锡便宜三至五倍而且我们比他们快三到五倍。

考虑到难找到数据科学家，值得一提的是，虽然有这么多选择性，但是Experfy是有其新员工培训的过程。显然,数据科学家们一个敏感而且容易满足的无聊如果左一遍又一遍地做同样的事情。这可以解释为什么很多人后悔进入Experfy的系统，以及为什么Experfy能够这么好挑剔的。

“数据科学想要的人是一定程度的智力上的开发。如果他们被困在一个朝九晚五的工作，他们不是很高兴”。因为Experfy提供他们的环境中有很多不同的问题他们可以解决。”

Singh说，有四个主要类型的数据科学家被Experfy所吸引：

• 自由职业者不满足，甚至认为能朝九晚五工作；
• 慢吞吞的人想通过兼职来增添氛围；
• 学者希望看到他们的算法应用于真实的世界；

  精品数据科学公司正在试图建立一个客户基础。

Experfy在一年多的时间内，比其它创业公司有更大的增长速度。Singh的十几个员工是他在哈佛创新实验室的人，并将可能需要雇佣更多的科学大数据需求升温。“这只是冰山的一角，”他说。“大多数公司还没有接受分析；只有财富500强的才有深刻体验。但是每个人都可以使用分析。每个人都可以是数据驱动的，每个公司都可以通过数据才做决策与支持，从而创造更大的价值。”

英文全文：

Starting a big data analytics project can be tough, especially for mid-size firms that can’t justify the cost of a full-time data scientist and don’t want to deal with traditional consulting firms. But a company called Experfy is finding success by emulating the sharing-economy model popularized by Uber, and matching freelance data scientists with analytic projects of clients.

Experfy came out of the Harvard Innovation Lab about a year ago with the goal of creating a marketplace for high-end consulting work revolving around data analytics. To date, the company has attracted a stable of 1,200 data scientists, who create proposals and bid on data projects submitted by clients.

It’s another sign that the sharing economy has come to data science. “We’re the Uber of data science,” says Harpreet Singh, co-founder and co-CEO of Experfy. “If you’re a medium-size business in the U.S. and you want to do data analysis, you can hire somebody or go to a consulting firm. But consulting firms are very expensive, and hiring somebody on the payroll only makes sense if you’re going to have a recurring need.”

So far, Experfy’s data science freelancers have successfully completed about 60 projects. The projects themselves run the gamut from creating equity trading algorithms and building recommendation systems in Hadoop, to translating Excel models into software as a service (SaaS) apps and creating dashboards in Tableau.

A sample of data analytics jobs currently posted on Experfy’s website

In some cases, Experfy’s clients just want an answer to a data-oriented question, while in other cases, they desire the algorithm, expressed in R or Python, which they’ll implement on their own. Some data scientists working for Experfy may serve temporarily as project managers, while teams of data scientists may join a client’s advisory board to provide recommendations and review technical proposals by third party vendors.

“We’re doing a lot of handholding with the clients,” Singh tells Datanami. “Analytics is a complicated thing. You can’t just go post a project and the magic happens. There’s a lot of hard work involved. The clients appreciate that. They want someone who will work with them and help them through the process.”

Experfy works with the clients to craft the description of their project before posting it to the online marketplace. Once it’s online, Experfy’s data scientists can submit their own proposals. “Just reading the proposals is a real treat for the clients because they learn so much about their problem, the different perspectives,” Singh says.

After considering the proposals, Experfy helps the client narrow it down to the two or three best ones, and the parties engage in videoconference calls on the Experfy site. Once the winning bid is selected, Experfy works with the two sides to ensure the work is done on time and meets expectations. Payments are kept in escrow until the client is satisfied. Experfy, which charges a 20 percent commission, guarantees success; no projects have failed, Singh says.

To increase the odds of success, Experfy encourages its clients to post smaller data science projects engagements. “Longer engagements, bigger projects, generally are riskier,” Singh says. The average Experfy engagement lasts two to three weeks and costs $10,000 to $20,000. However, there’s no limit; one recent project to build an IBM Watson-powered application took several months and cost $300,000.

It’s all about making data science accessible to a new class of businesses. “It becomes very affordable for somebody to say, ‘I have this question or need to make this decision’, or ‘I’ve got a business problem and need an algorithm that can solve that problem.’ The data scientists can easily come and solve that problem without a lot of risk,” Singh says. “We’re three to five times cheaper than McKinsey, but we’re three to five times faster as well.”

Experfy closely vets each data scientist who applies to the program. About 80 percent of the data scientists who apply are denied; most of them are rejected by the powerful algorithms that Experfy uses to profile the data scientists from their public profiles available in places like GitHub, Stack Overflow, and Kaggle. But some are rejected by Experfy’s administrators, too.

Considering how difficult it is to find data scientists (i.e. “unicorns”), it’s remarkable how selective Experfy is with its onboarding process. Apparently, data scientists are a sensitive lot who are prone to fits of boredom if left doing the same thing over and over. That could explain why so many are angling to get into Experfy’s system, and why Experfy can afford to be so choosey.experfy logo

“The people who do data science want a degree of intellectual stimulation. If they’re stuck in a 9 to 5 job, they’re not very happy,” Singh says. “What Experfy provides them is an environment where there are a lot of different problems they can solve.”

Singh says there are four main types of data scientists who are attracted to Experfy:

Freelancers who are too restless to even consider 9-5 jobs;

The 9-5ers who want to spice things up by moonlighting on the side;

Academics who desire to see their algorithms used in the real world;

Boutique data science firms are trying to build a customer base.

Experfy has been around for just over a year, and is growing at the ridiculous clip that startups do. Singh has about a dozen employees working for him in the Harvard Innovation Lab, and will likely need to hire more as demand for big data science heats up. “This is just the tip of the iceberg,” he says. “Most companies haven’t yet embraced analytics; forget about big data–that’s just for the Fortune 500 perhaps. But everybody can use analytics. Everybody can be data driven.”

End.