押宝大数据

    去年，企业家出身的风险投资家维诺德•科斯拉曾表示，有朝一日，科技将取代80%的医生，这个说法掀起轩然大波。日前，在Ayasdi公司于加州门洛帕克市举办的一次活动中，科斯拉再度表示，计算能力——而非人类——将给几乎所有领域带来巨大改善，尤其是在医疗保健领域。Ayasdi是科斯拉的公司投资的一家企业。

    科斯拉对听众说：“医生掌握的几乎所有医学知识都将被淘汰。”今年早些时候，资深技术专家、太阳微系统公司（Sun Microsystems）联合创始人科斯拉牵头对Ayasdi进行了一轮1000万美元的投资。Ayasdi是一家成立不久的大数据公司，利用“拓扑分析”在各种海量数据中寻找模式。（简言之，Ayasdi将海量线性数据，比如在医疗档案中找到的数据，转变成人们可以与之互动的几何形状。）Ayasdi位于加州帕洛阿尔托，是众多大数据新创企业中的一家。不过，大多数大数据公司都是依靠数据科学家向系统提问，而Ayasdi则声称，它的技术能使企业用户收集到意想不到的数据。Ayasdi的早期客户包括通用电气（General Electric）、花旗（Citi）以及默克公司（Merck）等。

    科斯拉说：“在高度复杂的数据中，最有意思的其实是我们想不到要问的问题。”Ayasdi吸引科斯拉的地方是它能够使用更多的计算能力，而非数据科学家，来攀登数据科学的高峰。数据科学家都是抢手的香饽饽，但不同的是，计算能力的成本事实上正在降低。

    Ayasdi还抓住了另一种趋势，那就是数据可视化。从位于西雅图的Tableau到帕洛阿尔托的ClearStory Data这样刚刚创建的公司，大大小小的企业都在提供工具，帮助用户将海量数据转化成漂亮的交互式图表。技术研究公司高德纳（Gartner）称，大数据这个产品类别——包括数据可视化以及其它工具——今年将吸引340亿美元的IT开支。购买大数据工具的企业可谓形形色色，它们的多元化程度不逊色于我们所分析的大数据，所涉及的领域从医疗保健到零售再到职业体育可谓无所不包。

    不过，至少还有一个区域还没怎么受到大数据的影响，那就是风险投资界。科斯拉本人就是一名资深的风险投资家，多年来进行的投资已经超过一百笔。但他表示，自己从来没有计算过投资回报率，也没有使用过任何数据科学工具来确定某笔投资可能的结果。

    关于使用大数据来为帮助自己进行投资决策，科斯拉说：“（这方面）没有足够大的数据库。”目前尚不清楚科斯拉对Ayasdi的投资会得到什么样的结果，也不清楚他对医疗保健的未来预测是否准确。所以，就算拓扑数据分析也并不是无所不能。（财富中文网）

    译者：项航     

    Last year, entrepreneur-turned-venture capitalist Vinod Khosla made waves when he said technology would someday replace 80% of doctors. This morning, at a Menlo Park, Calif.-based event hosted by one of his firm's portfolio companies, Ayasdi, Khosla reiterated his belief that computational power--not people--will bring about massive improvements in nearly every field, particularly healthcare.

    "Almost everything doctors know about medicine will be obsolete," Khosla told the audience. Earlier this year, the long-time technologist and co-founder of Sun Microsystems led a $10 million funding round in Ayasdi, a newish big data company that utilizes "topological analysis" to look for patterns in massive, often disparate data points. (In plain English, Ayasdi turns mountains of linear information, like what is found in health records, into geometric shapes that people can interact with.) The Palo Alto, Calif.-based company is one of an exploding number of big data start-ups. But while most companies' offerings rely on data scientists to query systems with questions, Ayasdi says its approach enables companies to glean insights they didn't know they were looking for. General Electric, Citi and Merck are among the company's early customers.

    "The most interesting things in highly complex data are the things we don't know to ask," says Khosla, who was attracted to Ayasdi's ability to scale data science using more computational power, not more data scientists. Unlike coveted data scientists, compute power is actually getting cheaper.

    But Ayasdi is also tapping into another trend: data visualization. From Seattle-based Tableau to just-launched startups like Palo Alto's ClearStory Data, companies large and small are offering up tools that help users spin beautiful, interactive graphics from massive data sets. According to technology research firm Gartner, the big data category--visualization and other tools--will drive $34 billion of IT spending this year. The companies who are buying up big data tools are almost as diverse as the disparate data sets we can now analyze, from healthcare to retail to professional sports.

    At least one area that big data has yet to significantly affect is the venture capital world. Khosla has made more than 100 investments in his many years as a venture capitalist, but says he has never calculated ROI or used any kind of data science tools to determine the likely outcome of an investment.

    "There is no database large enough," he says on the problem of using big data to inform his investing decisions. It remains to be seen how his bet on Ayasdi--or predictions on the future of healthcare--will pan out. Even topological data analysis has its limitations.