{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"color","attrs":{"color":"#333333","name":"user"}},{"type":"strong"}],"text":"本文最初发布于Uber工程博客，由InfoQ中文站翻译并分享。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"前言"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据对于我们的产品而言至关重要。数据分析帮助我们为使用我们服务的用户提供了流畅的体验。它也让工程师、产品经理、数据分析师、数据科学家可以在了解情况后作出明智的决定。数据分析影响了App的每一个界面：在主界面上显示什么，产品以什么顺序显示，向用户显示哪些相关的信息，什么妨碍了用户乘车或注册，诸如此类。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如此大的用户群体，如此广泛的特性，还要覆盖所有的地理区域，这是一个很复杂的问题。而且，我们的App一直在推出新产品，这就要求底层的技术也要有足够的灵活性来支持这种发展。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据是实现这种发展的最基本工具。本文将聚焦乘客数据：我们如何收集和处理以及这些数据具体如何影响了乘客端App的改进。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"乘客数据"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"乘客数据包含了乘客与Uber乘客端App的所有交互。其中每天都会有来自Uber在线系统的数十亿个事件，这些事件转换成了数百张Apache Hive 表，为乘客端App的不同应用场景提供支持。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"下面是可以利用乘客数据分析的主要问题领域："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"增加漏斗转化"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"提高用户参与度"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"个性化"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"用户沟通"}]}]}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"在线数据收集"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"移动事件日志"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"乘客数据有多个来源，但最基本的一个是获取用户与App的交互过程。用户交互是通过移动端的事件日志获取的。下面是日志架构设计的一些关键原则："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"日志标准化"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"跨平台一致性（iOS、Android、Web）"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"尊重用户隐私设置"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"优化网络使用"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"可靠但不降低用户体验"}]}]}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"日志标准化"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有一个标准的日志记录过程很重要，因为数以百计的工程师在增加或编辑事件。从客户端收集的日志有的是平台化的（像用户与UI元素的交互事件、内容曝光次数等），有的是由开发人员手动添加的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们将一组元数据进行了标准化，作为默认的公共负载随每个事件发送，如位置、App版本、设备、暱称等。这对于后台指标计算至关重要。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此外，为了确保所有事件跨所有平台都能保持一致，并且有标准的元数据，我们定义了Thrift结构，事件模型需要实现这个结构来定义其有效负载。Thrift模式包含一个枚举（表示在不同平台上的事件ID）和一个有效负载结构（定义了事件注册时与之关联的所有数据），最后还有一个事件类型。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b8\/56\/b8c5b31b70fee25104247b0f33173e56.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"示例：Thrift 模式中分析事件的标准化定义"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"发布日志"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这些日志通过管道进入Unified Reporter，这是客户端里的一个框架，用于摄取客户端产生的所有消息。Unified Reporter会将消息存储在队列中，对它们进行聚合，然后通过网络每隔几秒分批次地发送给后台的Event Processor。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/ac\/f4\/ac089866bebbe0249d2f92407f2158f4.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"图1 事件被记录到仪表盘和数据集的过程"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"事件一直在增加或变化——每天处理的事件有几百种类型。其他日益严重的问题还有：跨不同操作系统（Android和iOS）的日志平台化、可发现性以及如何保持良好的信噪比。Event Manager门户负责管理这些事件的元数据，并为事件选择合适的接收器。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Event Processor根据接收到的元数据确定如何处理事件以及进一步传播。此外，如果事件的元数据和映射不可用，Event Processor就会阻挡该事件，不再向下游传播。这是为了提升信噪比。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"后台事件日志"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"伴随用户交互，我们要记录App向用户展示了什么内容，这很重要。我们是通过在后台记录服务层的数据来实现的。后台日志记录处理的数据更多，有些是移动端没有的，有些是移动端处理不过来的。由移动端或其他系统发起的每次后端调用都会有数据记录。每条记录都有一个”join“键，通过它可以关联到移动端交互。这项设计可以保证移动端带宽得到有效使用。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"离线数据处理"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们把从移动端和服务层收集到的数据进行结构化，并作为离线数据集进行复制。离线数据集帮助我们识别上文提到的问题，并评估为解决这些问题所开发的解决方案有多成功。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"原始的大型离线数据集真得很难处理。我们对原始数据进行扩充并建模，形成分层表。在扩充过程中，我们把不同的数据集连接在一起，让数据更有意义。建模形成的表可以带来以下几个方面的好处："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":null,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"节省资源"},{"type":"text","text":"：仅计算一次并存储。其他任何人都不需要在原始的大型数据集上运行查询。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"标准化定义"},{"type":"text","text":"：业务逻辑和指标定义都在ETL中（提取、转化、加载），不需要消费者操心。如果把这项工作留给消费者，那么每个团队可能会做不同的计算。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"数据质量"},{"type":"text","text":"：可以保证适当的检查对比，因为逻辑都在一个地方，数据很容易检验。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":4,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"所有权"},{"type":"text","text":"：随着数据演化，数据所有者可以确保表能够适用于新特性。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/4e\/db\/4e51354c9c9002ce7b562f943ee551db.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"图2 各种离线数据处理场景"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"让我们考虑一下下面这个问题描述："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":null,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":"快捷乘车改善了乘客体验，促成了更多转化（出行）吗？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们从保存了用户交互和主界面内容的基础事实表中筛选出与“快捷乘车”相关的信息，并通过与其他多个数据集集成对它进行扩充，进而实现漏斗分析："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户显示了快捷乘车区域？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户点击了其中的一个快捷方式？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户（来自#2）最终预定了出行？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户（来自#3）完成了出行？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"通过快捷乘车流程和普通流程完成出行的主界面曝光比是多少？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"快捷乘车对于出行预定的总体效果是什么？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"2."},{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":" 奖励计划对于乘客的作用有多大？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"为了找出这个问题的答案，表中应该包含如下数据："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"选择\/兑换的奖励"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"未使用或过期的奖励"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"乘客如何赢得奖励？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"还有其他一些有趣的数据点，如："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"奖励计划增加了App的总体使用量吗？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支出是否与这项计划的预算相符？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/www.uber.com\/in\/en\/u\/rewards\/","title":null,"type":null},"content":[{"type":"text","text":"奖励"}]},{"type":"text","text":"可以通过Eats、Rides和其他Uber应用的不同功能进行兑换。一旦用户在移动端选择了一项奖励，就会触发中心化的奖励后端服务。它会处理奖励信息，将每个奖励选择行为记录为交易数据。有些奖励是自动应用的，但有些是促销驱动的。促销驱动的奖励兑换是在另一个促销系统中处理的。此外，这个系统的构建让运营或产品团队能够根据需要轻松添加新的奖励。我们构建工具来获取奖励元数据，这些元数据反过来又流向另一个系统。有个ETL作业会读取流经不同系统的数据，生成一个奖励兑换数据模型。此外，这些数据还能帮助财务团队获取Uber在奖励计划中的开销，让人们对这个产品有一个良好的理解。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"3."},{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":" 在COVID-19之后，Uber航空出行的恢复率是多少？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"航空出行的不同指标是从上游多个表收集的，包括出行、会议、理财、航空出行及其他乘客表等不同的领域。来自不同领域的数据会被聚合，然后在一组维度下计算成指标，存储到一个表中。将最新数据与经过聚合的历史数据进行比较，可以帮助我们找到上面这个问题的答案。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此外，航空出行数据还被用于绘制机场接机的落地数据热图，计算机场总接机量、总预定量等。所有这些数据都有助于我们业务的发展，还有助于业务本地化，满足不同地域的不同需求。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"数据质量"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据可以为我们提供业务决策的依据。因此，保证数据的完整性和质量变得非常重要。在乘客端App的架构中，为了保证数据质量，我们在多个层面做了数项检查。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在产生事件的时候，我们引入了测试框架进行构建时测试、模式和语义检查。这些框架会检查是否有分析事件被触发，有效负载、顺序是否符合预期。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b5\/f3\/b5e5f4600bca82d63ca71e28d4767ef3.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"图3 数据流数据质量检查"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"一旦事件到达离线存储并处理，异常检测功能就可以保证数据被记录并按照预期流转。系统会监控事件量，如果突然出现下降或峰值，就给所有者发送告警信息。这种监控有助于捕捉差异，防止出现中断而没有发现。在离线建模的表中，测试框架被用于确保数据的正确性、覆盖率以及各表之间的一致性。每次管道运行都会触发配置好的测试，保证产生的任何数据都能满足质量SLA（服务水平协议）。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"Uber乘客端App的演进"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"根据上面这些从数据收集机制中了解到的东西，我们对乘客端App做了一些更改，下面是几个具体的例子："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b7\/44\/b73dc73c7fe949a2a1c55fyyc9f31a44.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"并非所有特性在所有市场中都可用"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"高质量的数据是推动应用程序演进的强大工具。不说别的，它可以帮助我们改善用户体验，这反过来又增加了用户粘度，促进了用户增长。此外，在添加新特性的时候，数据可以告诉我们什么最适合用户，保证更改不会导致用户体验下降。我们深刻理解数据的重要性，"},{"type":"link","attrs":{"href":"https:\/\/eng.uber.com\/ubers-journey-toward-better-data-culture-from-first-principles\/","title":null,"type":null},"content":[{"type":"text","text":"我们一直在提升Uber的数据文化"}]},{"type":"text","text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"查看英文原文："},{"type":"link","attrs":{"href":"https:\/\/eng.uber.com\/how-data-shapes-the-uber-rider-app\/","title":null,"type":null},"content":[{"type":"text","text":"How Data Shapes the Uber Rider App"}]}]}]}
Uber如何处理和使用乘客数据改善App的体验？
{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"color","attrs":{"color":"#333333","name":"user"}},{"type":"strong"}],"text":"本文最初发布于Uber工程博客，由InfoQ中文站翻译并分享。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"前言"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据对于我们的产品而言至关重要。数据分析帮助我们为使用我们服务的用户提供了流畅的体验。它也让工程师、产品经理、数据分析师、数据科学家可以在了解情况后作出明智的决定。数据分析影响了App的每一个界面：在主界面上显示什么，产品以什么顺序显示，向用户显示哪些相关的信息，什么妨碍了用户乘车或注册，诸如此类。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如此大的用户群体，如此广泛的特性，还要覆盖所有的地理区域，这是一个很复杂的问题。而且，我们的App一直在推出新产品，这就要求底层的技术也要有足够的灵活性来支持这种发展。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据是实现这种发展的最基本工具。本文将聚焦乘客数据：我们如何收集和处理以及这些数据具体如何影响了乘客端App的改进。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"乘客数据"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"乘客数据包含了乘客与Uber乘客端App的所有交互。其中每天都会有来自Uber在线系统的数十亿个事件，这些事件转换成了数百张Apache Hive 表，为乘客端App的不同应用场景提供支持。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"下面是可以利用乘客数据分析的主要问题领域："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"增加漏斗转化"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"提高用户参与度"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"个性化"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"用户沟通"}]}]}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"在线数据收集"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"移动事件日志"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"乘客数据有多个来源，但最基本的一个是获取用户与App的交互过程。用户交互是通过移动端的事件日志获取的。下面是日志架构设计的一些关键原则："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"日志标准化"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"跨平台一致性（iOS、Android、Web）"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"尊重用户隐私设置"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"优化网络使用"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"可靠但不降低用户体验"}]}]}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"日志标准化"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有一个标准的日志记录过程很重要，因为数以百计的工程师在增加或编辑事件。从客户端收集的日志有的是平台化的（像用户与UI元素的交互事件、内容曝光次数等），有的是由开发人员手动添加的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们将一组元数据进行了标准化，作为默认的公共负载随每个事件发送，如位置、App版本、设备、暱称等。这对于后台指标计算至关重要。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此外，为了确保所有事件跨所有平台都能保持一致，并且有标准的元数据，我们定义了Thrift结构，事件模型需要实现这个结构来定义其有效负载。Thrift模式包含一个枚举（表示在不同平台上的事件ID）和一个有效负载结构（定义了事件注册时与之关联的所有数据），最后还有一个事件类型。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b8\/56\/b8c5b31b70fee25104247b0f33173e56.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"示例：Thrift 模式中分析事件的标准化定义"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"发布日志"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这些日志通过管道进入Unified Reporter，这是客户端里的一个框架，用于摄取客户端产生的所有消息。Unified Reporter会将消息存储在队列中，对它们进行聚合，然后通过网络每隔几秒分批次地发送给后台的Event Processor。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/ac\/f4\/ac089866bebbe0249d2f92407f2158f4.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"图1 事件被记录到仪表盘和数据集的过程"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"事件一直在增加或变化——每天处理的事件有几百种类型。其他日益严重的问题还有：跨不同操作系统（Android和iOS）的日志平台化、可发现性以及如何保持良好的信噪比。Event Manager门户负责管理这些事件的元数据，并为事件选择合适的接收器。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Event Processor根据接收到的元数据确定如何处理事件以及进一步传播。此外，如果事件的元数据和映射不可用，Event Processor就会阻挡该事件，不再向下游传播。这是为了提升信噪比。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"后台事件日志"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"伴随用户交互，我们要记录App向用户展示了什么内容，这很重要。我们是通过在后台记录服务层的数据来实现的。后台日志记录处理的数据更多，有些是移动端没有的，有些是移动端处理不过来的。由移动端或其他系统发起的每次后端调用都会有数据记录。每条记录都有一个”join“键，通过它可以关联到移动端交互。这项设计可以保证移动端带宽得到有效使用。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"离线数据处理"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们把从移动端和服务层收集到的数据进行结构化，并作为离线数据集进行复制。离线数据集帮助我们识别上文提到的问题，并评估为解决这些问题所开发的解决方案有多成功。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"原始的大型离线数据集真得很难处理。我们对原始数据进行扩充并建模，形成分层表。在扩充过程中，我们把不同的数据集连接在一起，让数据更有意义。建模形成的表可以带来以下几个方面的好处："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":null,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"节省资源"},{"type":"text","text":"：仅计算一次并存储。其他任何人都不需要在原始的大型数据集上运行查询。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"标准化定义"},{"type":"text","text":"：业务逻辑和指标定义都在ETL中（提取、转化、加载），不需要消费者操心。如果把这项工作留给消费者，那么每个团队可能会做不同的计算。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"数据质量"},{"type":"text","text":"：可以保证适当的检查对比，因为逻辑都在一个地方，数据很容易检验。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":4,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"所有权"},{"type":"text","text":"：随着数据演化，数据所有者可以确保表能够适用于新特性。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/4e\/db\/4e51354c9c9002ce7b562f943ee551db.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"图2 各种离线数据处理场景"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"让我们考虑一下下面这个问题描述："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":null,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":"快捷乘车改善了乘客体验，促成了更多转化（出行）吗？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们从保存了用户交互和主界面内容的基础事实表中筛选出与“快捷乘车”相关的信息，并通过与其他多个数据集集成对它进行扩充，进而实现漏斗分析："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户显示了快捷乘车区域？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户点击了其中的一个快捷方式？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户（来自#2）最终预定了出行？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有多少用户（来自#3）完成了出行？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"通过快捷乘车流程和普通流程完成出行的主界面曝光比是多少？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"快捷乘车对于出行预定的总体效果是什么？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"2."},{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":" 奖励计划对于乘客的作用有多大？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"为了找出这个问题的答案，表中应该包含如下数据："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"选择\/兑换的奖励"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"未使用或过期的奖励"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"乘客如何赢得奖励？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"还有其他一些有趣的数据点，如："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"奖励计划增加了App的总体使用量吗？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"支出是否与这项计划的预算相符？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/www.uber.com\/in\/en\/u\/rewards\/","title":null,"type":null},"content":[{"type":"text","text":"奖励"}]},{"type":"text","text":"可以通过Eats、Rides和其他Uber应用的不同功能进行兑换。一旦用户在移动端选择了一项奖励，就会触发中心化的奖励后端服务。它会处理奖励信息，将每个奖励选择行为记录为交易数据。有些奖励是自动应用的，但有些是促销驱动的。促销驱动的奖励兑换是在另一个促销系统中处理的。此外，这个系统的构建让运营或产品团队能够根据需要轻松添加新的奖励。我们构建工具来获取奖励元数据，这些元数据反过来又流向另一个系统。有个ETL作业会读取流经不同系统的数据，生成一个奖励兑换数据模型。此外，这些数据还能帮助财务团队获取Uber在奖励计划中的开销，让人们对这个产品有一个良好的理解。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"3."},{"type":"text","marks":[{"type":"italic"},{"type":"strong"}],"text":" 在COVID-19之后，Uber航空出行的恢复率是多少？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"航空出行的不同指标是从上游多个表收集的，包括出行、会议、理财、航空出行及其他乘客表等不同的领域。来自不同领域的数据会被聚合，然后在一组维度下计算成指标，存储到一个表中。将最新数据与经过聚合的历史数据进行比较，可以帮助我们找到上面这个问题的答案。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此外，航空出行数据还被用于绘制机场接机的落地数据热图，计算机场总接机量、总预定量等。所有这些数据都有助于我们业务的发展，还有助于业务本地化，满足不同地域的不同需求。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"数据质量"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据可以为我们提供业务决策的依据。因此，保证数据的完整性和质量变得非常重要。在乘客端App的架构中，为了保证数据质量，我们在多个层面做了数项检查。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在产生事件的时候，我们引入了测试框架进行构建时测试、模式和语义检查。这些框架会检查是否有分析事件被触发，有效负载、顺序是否符合预期。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b5\/f3\/b5e5f4600bca82d63ca71e28d4767ef3.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"图3 数据流数据质量检查"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"一旦事件到达离线存储并处理，异常检测功能就可以保证数据被记录并按照预期流转。系统会监控事件量，如果突然出现下降或峰值，就给所有者发送告警信息。这种监控有助于捕捉差异，防止出现中断而没有发现。在离线建模的表中，测试框架被用于确保数据的正确性、覆盖率以及各表之间的一致性。每次管道运行都会触发配置好的测试，保证产生的任何数据都能满足质量SLA（服务水平协议）。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"Uber乘客端App的演进"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"根据上面这些从数据收集机制中了解到的东西，我们对乘客端App做了一些更改，下面是几个具体的例子："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b7\/44\/b73dc73c7fe949a2a1c55fyyc9f31a44.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"并非所有特性在所有市场中都可用"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"高质量的数据是推动应用程序演进的强大工具。不说别的，它可以帮助我们改善用户体验，这反过来又增加了用户粘度，促进了用户增长。此外，在添加新特性的时候，数据可以告诉我们什么最适合用户，保证更改不会导致用户体验下降。我们深刻理解数据的重要性，"},{"type":"link","attrs":{"href":"https:\/\/eng.uber.com\/ubers-journey-toward-better-data-culture-from-first-principles\/","title":null,"type":null},"content":[{"type":"text","text":"我们一直在提升Uber的数据文化"}]},{"type":"text","text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"查看英文原文："},{"type":"link","attrs":{"href":"https:\/\/eng.uber.com\/how-data-shapes-the-uber-rider-app\/","title":null,"type":null},"content":[{"type":"text","text":"How Data Shapes the Uber Rider App"}]}]}]}
發表評論
所有評論
還沒有人評論，想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章