Facebook大数据训练营的启示

    大家可能早就听说过Facebook公司的工程师训练营。这是一个为期六周的上岗培训项目，主要是让新员工深入学习公司的代码库，同时了解公司文化。过去几个月里，这个社交网络巨头还悄悄地推出了另外一个项目，它不是专门面向工程师的——而是要教所有员工学会使用大数据工具。

    Facebook的数据分析部门主管肯•鲁丁说：“我们的确希望每个人都觉得自己有能力运用数据。这样一来，数据分析师（又称数据达人）就不会成为完成任务的瓶颈了。分析师是专门待命去执行特种部队式的任务的，就是那些不管规模，还是深度都非一般人所能企及的任务。”

    Facebook旗下约有100名这类分析师（并且分析师队伍还有大量职位正虚位以待）。不过，曾任社交游戏公司Zynga分析师与平台技术副总的鲁丁表示，他想在公司推广这样一种文化，使每个人都能用数据测试、并最终推出新产品、设计修改和其他改进。为此，鲁丁和他的团队已经尝试推出了各种教授如何运用数据分析工具的辅导课程。去年11月，他们推出了首个为期两周的大数据课程，现在正在持续不断地授课。而后续还将推出各种课程。每次为期两周的课程最多能容纳25名员工——包括产品经理，客服人员，公司基础架构团队成员。他们在这两周里每天都要上课，上午是三小时集中授课，下午完成自选项目。每个学员这两周中都会有一位专职导师辅导，公司希望他们每个人都能解决一个公司面临的实际问题（比如如何运用数据提供更好的客户服务）。

    Facebook本来就一直靠数据进行决策，同时依靠真实用户来测试新产品。它开发了自己的大数据工具，帮助所有员工——而不仅仅是分析师——方便快速地在其它庞大的数据库中进行查询。比如，HiPal就是一个旨在让公司所有员工都能方便地分析高达拍字节数据的工具。Gatekeeper是另一个工具，管理Facebook每天开展的成百上千个用户测试项目，它能确保这些测试提供“从统计角度而言有意义的结果。”

    不过鲁丁强调，开展这种培训绝不仅仅是为了推广合适的工具——它的根本目的还在于培养员工的思维能力。公司的大数据训练营负责教员工如何进行探索性分析并得出假设，还训练他们有效沟通并演示自己的分析成果。鲁丁说：“我认为，如果我们坚持现在的做法，最后就能达成目标，那样就能培养一种特有的公司文化，每个人都会感到数据是他们开展工作必须运用的一部分。每个人都应该具有数据分析的能力。”

    当然，鲁丁和他的团队自己也正在运用数据弄清楚如何改进下一届训练营——哪些课程最有效，如何才能确定课程的最佳规模。要找到天赋出众的分析师本身就够困难的了（更别说高昂的费用），而要让Facebook近5,000名员工自愿参加数据分析培训也不那么容易。其他公司会效法吗？对Facebook的公司文化来说，这种两周的强化课程很有意义，因为这种训练营式的项目早就是新晋工程师的必经仪式。不过，训练员工学会运用大数据工具——同时培养数据分析的意识，对其他公司来说也颇有裨益。（财富中文网）

    译者：清远

    You may have heard of Facebook's engineering bootcamp, a six-week onboarding program for new hires to learn the ins and outs of the company's code base and culture. But over the last few months, the social networking giant has quietly rolled out another program that's not just for engineers -- rather, it's focused on teaching big data tools to all employees.

    "We really want everyone to feel like they are capable of using data," says Ken Rudin, head of analytics at Facebook (FB). "Then analysts [a.k.a. data crunchers] aren't a bottleneck to getting things done. They're there for doing the SWAT team type of things, things that take a little extra scale and more depth than your average person would have."

    Facebook employs about 100 so-called analysts (and lists plenty of open positions for its analytics team). But Rudin, formerly VP of analytics and platform technologies at Zynga (ZNGA), says he wants to promote a culture in which everyone uses data to test and ultimately roll out new products, design changes, and other improvements. To that end, Rudin and his team have experimented with different kinds of tutorial sessions on using data analytics tools. Last November, they launched the first two-week session on big data and are now running courses back to back; there is a wait-list for upcoming sessions. Each two-week course consists of up to 25 employees -- product managers, customer service workers, and members of the company's infrastructure team, for example. They come in every day for two weeks, sitting in on about three hours of lectures each morning and then taking the rest of the day to work on self-selected projects. Each employee is assigned a mentor for the duration of the two weeks and is expected to work on a real company problem (such as how to use data to provide better customer service).

    Facebook has a history of data-driven decisions and running tests on real users to try out new products. The company has developed homegrown big data tools to help all sorts of employees -- not just analysts -- quickly and easily run queries on its immense data sets. HiPal, for example, is a tool that aims to make analyzing petabytes of data easy for anyone in the company. Gatekeeper is another tool that manages the hundreds of user tests Facebook runs each day and makes sure that they provide "statistically meaningful results."

    But Rudin stresses that it's not just about having the right tools -- it's about the right mindset. The company's big data bootcamp teaches employees how to conduct exploratory analysis and come up with hypotheses. It also trains them to effectively communicate and present their findings. "If we continue down the path that we're going, and I think we'll get there, then we'll have a culture where everyone feels that data is something they should be using as part of their job," says Rudin. "Everybody should be doing analysis."

    Of course, Rudin and his team are also using data to figure out how to evolve their new bootcamp -- what type of curriculum is most effective and how they can best scale the courses. While finding talented analysts is hard (not to mention expensive), putting Facebook's nearly 5,000 employees through a voluntary boot camp on crunching numbers isn't easy either. So will other companies follow suit? A two-week intensive course makes sense for Facebook's culture, where a bootcamp-style program has become a rite of passage for incoming engineers. But other companies could benefit from training employees to adopt a big data toolset -- and mindset.