Big data cases 35-40: Data Association Analysis, Data exhaust, Dark data

Next data association, data exhaust and dark data

Big data is mainly not judged by causality and is mainly applicable to correlation analysis. Many correlation analyses do not require complex models and only require awareness of big data.

Many organizations have data exhaust. The data is not used up or discarded. Its reuse value may not be clear to you now, but at some point in the future, it will burst out and turn waste into treasure.

Dark data is the data that is collected for a single target. It is usually archived and idle after being used, and its true value is not fully exploited. If the dark data is used in the right place, the company's business can be brightened.

35 Data Association Analysis

A company team used location data from mobile phones to figure out how many people stopped at the Macy’s parking lot on the day of the Christmas shopping season, and could predict their sales for the day, much earlier than Macy’s. Self-stated sales records. Whether it's an analyst on Wall Street or an executive in a traditional industry, you'll gain a great competitive advantage with this keen insight.

For tax authorities, tax fraud is increasingly being watched, and big data can be used to increase the government's process of identifying fraud. Where privacy permits, government agencies can synthesize various aspects of data such as vehicle registration and overseas travel data to discover individual spending patterns so that tax contributions are not superimposed. At the same time, a suspicious problem has emerged. There is no direct evidence pointing to fraud. These conclusions cannot be used to sue individuals. But he can help government agencies to clarify their audits and other audits as well as some processes.

36 data exhaust

The logistics company's data originally only served the operational needs, but once reused, the logistics company turned into a financial company, the data used to assess the customer's credit, provide unsecured loans, or use the goods in transit as a mortgage to provide loans; logistics The company can even transform into a financial information service company to judge the operation and trend of each segment of the economic sector.

Some companies have already had a feeling of "God overlooking" in big data. A company in Los Angeles announced that they have modeled the historical data of the global night scene and made a research report on investment real estate and consumption after filtering out the fluctuations. McDonald's, through the delivery service, obtained the user's precise address while selling the burger. After the collection of these address data, it became a wonderful internal data of the real estate industry.

37 dark data

In certain cases, dark data can be used for other purposes. Infinity Property & Casualty used a cumulative claims adjuster report to analyze fraud cases and recovered $12 million in subrogation compensation through an algorithm. An electric sales company, through the accumulation of 10 years of ERP sales data analysis, in accordance with the life cycle of electrical equipment, to visit the old customers five years ago, received more than 10 million yuan electrical equipment maintenance orders, smoothly entered the MRO market.

38 Customer churn analysis

American Express can only achieve post-mortem reports and lagging predictions. Traditional BI has been unable to meet the needs of its business development. As a result, AmEx began to build models that truly predict customer loyalty, based on historical transaction data, using 115 variables for analysis and prediction. The company said it has been able to identify 24% of Australia's customers who will be lost in the next four months. Such customer churn analysis can of course be used to retain customers. The hotel industry can customize the corresponding unique personality room for consumers, and even put the consumer's Weibo travel mood on the wallpaper and so on. Tourism can rely on big data to provide consumers with local products, activities, small and beautiful niche attractions that they may like, to restore the hearts of tourists.

39 Video Analysis of the Fast Food Industry

Fast food companies can analyze the length of the queue by video analysis and then automatically change the content displayed on the electronic menu. If the queue is long, it shows food that can be quickly supplied; if the queue is short, it shows those foods that are more profitable but have a longer preparation time.

40 big data campaign

In 2012, the Obama team that participated in the campaign identified three fundamental goals: to allow more people to pay more, to allow more voters to vote for Obama, and to get more people involved! This requires “micro” Cognition at the level: What is the most likely to be persuaded by each voter? Under what circumstances is each voter most likely to pay for it? What kind of advertising channel can get the most effective target voters? For example, campaign chief commander Jim Messina It is said that in the entire campaign, the assumption that there is no data to support cannot exist.

In order to raise $1 billion in campaign money, Obama's data mining team has collected, stored, and analyzed large amounts of data over the past two years. They noticed that movie star George Clooney is very attractive to women between the ages of 40 and 49 on the West Coast of the United States: they are undoubtedly the group most likely to pay for their own dinner in Hollywood with Clooney and Obama. . Clooney raised millions of dollars in campaign funding for Obama at a fundraising banquet hosted by his home. Later, when the Obama team decided to look for a movie star with the same appeal on the East Coast, the data team found that fans of Sarah Jessica Parker also liked competitions, small parties and celebrities. The "Crony effect" was successfully replicated on the East Coast.

Throughout the campaign, the Obama team spent less than $300 million on advertising costs, while the Romney team lost nearly $400 million, one of the important reasons being that Obama’s data team bought for advertising. Decision making was made after careful data analysis. A poll shows that 80% of American voters think that Obama is more sensitive to himself than Romney. As a result, 98% of the first $100 million raised by the Obama team came from small donations of less than $250, while the Romney team raised 31% in the same amount of donations.