孟山都首先发起“Green Data Revolution”运动,建立农业数据联盟(Open Ag Data Alliance)来统一数据标准,让农民不用懂“高科技”也能享受大数据的成果。典型的应用如农场设备制造商John Deere与DuPont Pioneer当前联合提供“决策服务(Decision Services)”,农民只需在驾驶室里拿出平板电脑,收集种子监视器传来的数据,然后将其上传给服务器,最终服务器返回化肥的配方到农场拖拉机上。
Here’s an example. An MRI scan is the best way to see inside the human body. While effective in helping to diagnose multiple sclerosis, brain tumors, torn ligaments and strokes, the data produced by an MRI machine is disconnected from the person that needs it the most.
At a very simplistic level, there are many individuals working as a team to make the scan happen. A nurse administers medications or contrast agents that may be needed for the exam; an MRI technologist operates the scanner; and a radiologist identifies the imaging sequences to be used and interprets the images. This information is then given to the nurse, who then passes it to the primary doctor to review and take action accordingly. This is Big Data, but it is not making information more intelligent.
When the program, a remake of a BBC miniseries, was up for purchase in 2011 with David Fincher and Kevin Spacey attached, the folks at Netflix simply looked at their massive stash of data. Subscribers who watched the original series, they found, were also likely to watch movies directed by David Fincher and enjoy ones that starred Kevin Spacey. Considering the material and the players involved, the company was sure that an audience was out there.
交通:《车来了》通过分析公交车上GPS定位系统每天的位置和时间数据,结合时刻表预测出每一辆公交车的到站时间;WNYC开发的Transit Time NYC通过开源行程平台(Github:OpenTripPlanner和MTA )获取的数据将纽约市划分成2930个六边形,模拟出从每一个六边形中点到边缘的时间(地铁和步行,时间是上午九点),最终建模出4290985条虚拟线路。用户只需点击地图或者输入地址就能知道地铁到达每个位置的时间;实时交通数据采集商INRIX-Traffic的口号是(永不迟到!^^),通过记录每位用户在行驶过程中的实时数据例如行驶车速,所在位置等信息并进行数据汇总分析,而后计算出最佳线路,让用户能够避开拥堵。
金融:ZestFinance | Big Data Underwriting 是由是Google的前任 CIO,Douglas Merrill创立金融数据分析服务提供商,使用机器学习算法和大数据为放款者提供承保模式,旨在为那些个人信用不良或者不满足传统银行贷款资格的个人提供服务。公司使用分析模型对每位信贷申请人的上万条原始信息数据进行分析,只需几秒时间便可以得出超过十万个行为指标。目前违约率比行业平均水平低 60%左右。另外一个不得不提到的是风险管理先驱者FICO | Predictive Analytics, Big Data Analytics and FICO Credit Scores,通过大数据分析为银行和信用卡发卡机构、保险、医疗保健、政府和零售行业提供服务。FICO 信用分计算的基本思想是:把借款人过去的信用历史资料与数据库中的全体借款人的信用习惯相比较,检查借款人的发展趋势跟经常违约、随意透支、甚至申请破产等各种陷入财务困境的借款人的发展趋势是否相似。FICO 已经为三分之二的世界 100 强银行提供服务,提高了客户忠诚度和盈利率、减少欺诈损失、管理信贷风险、满足监管与竞争要求并快速获取市场份额。想了解更多的企业可以看看附录中《经济学人》的文章《Big data: Crunching the numbers》。