让你的数据分析直接了当
Getting Your Analytics Straight
让你的数据分析直接了当
Alexander Youngblood Soria
Predictive Intelligence, Machine Learning, Artificial Intelligence, Advanced Analytics.
关键词:智能预测,机器学习,人工智能,进阶分析
All these terms are gaining attention at major marketing, business, and analytics conferences. If you’re like most people, you probably walk away from these conferences feeling like your company is behind in its analytical capabilities. However, don’t let the chatter fool you. The reality is that many companies aren’t ready to digest the most advanced machine learning algorithms just yet – but that shouldn’t stop you from building a roadmap for how to get there:
以上的这些关键词获取了大部分市场营销,商务,以及数据分析会议的焦点。或许你会像大部分人一样,径直越过这些会议,觉得这些分析技能还离你公司很远。但是,不要让众所云云迷惑你,事实上只是大部分的公司不具备消化这些先进计算机学习算法的能力,并不意味着你要止步于摸索靠近这些技能的道路。
Step 1: Invest in People
第一步:花钱请对的人
Data scientists, representing that rare mix of analytical expertise and business acumen that many companies are looking for today, have never been in higher demand. Hiring one of these sought-after talents to join your company is the first step to building out your analytics capabilities. However, getting the right person “on the bus,” as leadership expertJim Collinsadvises in his book “Good to Great,” is difficult in itself.
数据分析师,代表着很少部分的人,这些人既是数据分析的专家,也有商务上敏锐的嗅觉,这些人是大部分公司都在寻找的,并且现在的需求是前所未有的高涨。为自己的公司聘请一个受市场欢迎的人才是塑造公司数据分析能力的第一步。当然,找到一个这样对的人并不是意见简单的事,领导力专家 Jim Collins 在他的书“从优秀到卓越”提到这招聘的难度。
The skillset required of a data scientist has evolved; not only are data scientists asked to be the creators and translators of data-driven analytics solutions for the business, they are also increasingly the conduit to technology teams. You want to look beyond purely technical skills on a resume and seek out a talent for communicating complex topics in a way that a business user can understand, as well as some understanding of the technology environments from which the data originates and in which the solutions will be deployed.
数据分析师不再是传统意义上,只懂分析数据,需要升级为掌握一套的技能,需要分析并解读数据,为解决商业难题提供方法,同时也需要跟技术团队对接。在浏览人才简历的时候,不单只是寻找懂相关技术技能的,还需要找能懂得复杂商务问题,能像业务同事一样去沟通的,同时也要能分析读取数据的,知道需要用什么方法来开展业务的人才。
ACTION: Look for data scientists with a diverse background.
举措:找数据分析师要找有复合背景的。
Step 2: Get Back to the Basics
In the business world, the answer isn’t always found in the newest, coolest advanced algorithms or sophisticated machine learning applications – opportunity almost always reveals itself in the basics. In analytics, the “basics” typically refers to descriptive, predictive, and prescriptive statistics. What’s the difference?TechTargetdefines them this way:
在商业世界里,答案不一定隐藏在最新潮先进的算法或者最复杂得计算机学习应用中——恰恰是隐藏在基础中,在数据分析中,“基础”一词恰恰指的的描述性,预测性,指定的数据分析中。具体是什么呢? Tech Target给出以下的定义:
Descriptive analytics aims to provide insight into what has happened.
描述性指的是洞察出究竟发生了什么。
Predictive analytics helps model and forecast what might happen.
预测性指的是通过模型等预测将会发生什么。
Prescriptive analytics seeks to determine the best solution or outcome among various choices, given the known parameters.
指向性指的是在众多的方法中找到最佳的或者是效果最大化的方法,给出已知参数。
Think of descriptive, predictive, and prescriptive modeling as related tools in your data science toolbox; building off of these initial analytics techniques creates a solid foundation of knowledge to move to the next level in data science. That next level may very well involve more complex machine learning – but start with the basics.
综合以上三种维度出来的分析模型工具运用在你的数据分析中,通过这些最初的分析为数据分析进入下一阶段奠定坚实基础。下一阶段的数据分析将会涉及更加复杂得计算机学习技能——建立在以上基础上。
ACTION: Frame a problem in terms of descriptive analytics, which may very well answer 95 percent of a business stakeholder’s questions right away.
举措:通过描述性的分析来搭建问题框架,这基本上能直接解决一个投资人98%的疑问。
(真的做不完了,还有一堆工作没处理,下面简短【捂脸哭】)
Step 3: Organize Data in an Actionable Format
第三步:将数据输出变成一种可执行的部署
Data science is much more than cleaning and transforming data, running queries, and writing code, although there is a lot of that too. It’s about providing results back to your business stakeholders in a way that’s easily consumable. Spreadsheets full of numbers force business users to work through minute details of how a problem was solved when most of the time what they’re really looking for are answers and a recommendation to act. Even when the data science behind a solution is complex, you should consistently communicate the results in ways that simplify and summarize the solution in clear language. Even better, look for opportunities to deploy dashboarding and reporting software technologies to help automate and respond to some of the simpler, more commonly asked questions (e.g. what is our year-over- sales by product category?). This empowers your business stakeholders and frees up the data scientists to work on the hardest problems.
ACTION: Enable key business stakeholders to answer their own questions with data via dashboards and reporting.
举措:让数据分析的结果简单易懂,这样重要的投资者也能通过白板上的数据来进行商业汇报。
Step 4: Make Data Meaningful for the Business
第四步:让数据分析对商业更具有意义
We know it’s important for a data scientist to be an effective translator, finding the secrets hidden within the data and translating them into language that business users can understand and act upon. This deep dive into an understanding of customer behaviors lets marketers tailor their communications, interactions, and messaging in meaningful and relevant ways. Similarly, the data scientist should work hand in hand with the business to help it understand how industry trends and consumer preferences might change in the future, and to predict which customers may be affected most by a change.
ACTION: Integrate your data scientists tightly with the business so that they can learn to identify more relevant stories arising from the analytics, and anticipate key questions from stakeholders.
举措:让数据分析师紧密的融合在商业进行中,让他们能更好的理解数据发生的背景,甚至让他们来解答投资人的关键问题。
The industry is excited about the possibilities that predictive intelligence can unlock. With the rise of machine learning, everyone has an eye towards harnessing the power of advanced analytics to take the guesswork out of predicting the future. While there is no doubt that these complex capabilities are change agents for how marketers and business strategists will generate insights with data in the long term, don’t overlook the value of deploying your data scientists to solve problems with more straightforward analytics – the basics. As an acquaintance of mine puts it, “Don’t be the science fair in the back room – provide actionable insights for the business.” This can only be done by getting your analytics straight.