12 more big data cases
Habr, hello. Today, we have prepared 12 more examples of how big data technologies bring money to companies.
Industry: Telecom.
Beeline has a huge set of data about its subscribers, which the company plans to use not only for internal optimization and work with clients (sales increase, customer retention, anti-fraud), but also to bring new analytical products to the market (providing data for credit scoring, targeted digital advertising, creating geo-analytical reports, IPTV analytics, external consulting). The company has implemented many big data projects. For example, the subscriber base was segmented on the basis of an extended client profile, including with the prediction of gender and age category, the construction of social graphs; a project was implemented to recognize and protect subscribers from money frauds and viral activities; subscribers who use communication services on several types of devices, as well as subscribers located at the airport and flying abroad, were offered to offer them suitable services and tariff plans. And this is not a complete list. The company uses HDFS and Apache Spark for data storage and processing, rapidminer and python for data analysis, including the scikit-learn library.
Result: by 2018, the projected revenue from big data will be more than 20% of the company's revenue.
')
Industry: entertainment, gambling.
Caesars Entertainment is one of the leaders in the US gambling business. It operates the famous casino Caesar's Palace in Las Vegas and over 50 gambling establishments around the world. The company itself has recently been fined more than $ 20 million for money laundering, and its casino division has been subject to bankruptcy. However, for 17 years, the company has accumulated an impressive amount of data and analytics for its Total Rewards loyalty program, which is now one of its most valuable assets and was valued at $ 1 billion. In the gaming industry, data on each client is important: some players of the highest, seventh category according to the company's classification spent about $ 500,000 in casinos annually, and one businessman left $ 200 million in Las Vegas for the year. It is very important to be able to meet such a customer by name, hold on your favorite game and offer free additional options to his taste: entertainment, air tickets, limousine trips, hotel accommodation, etc. With the help of accumulated data on the tastes and behavior of customers and advanced methods of data analysis, a company could find such an approach to customers that they leave as much money as possible and come back again. The company uses Hadoop and cloud solutions, processing more than 3 million records per hour. Data analysis is also used to segment customers and improve security standards.
Result: the most valuable individual asset of the company worth more than $ 1 billion has been created, the growth of profitability and safety standards has been achieved.
Industry: dating sites.
eHarmony is a dating site focused on building long-term relationships. In the form on the site, the user can leave a very detailed information about yourself, specify more than 1000 parameters. Further, using these data, as well as the history of sympathies and the development of relations between the users of the site, the system recommends the most suitable dating options, and not just on the basis of similarity of interests and beliefs. User photos are analyzed, complementing each other and coinciding, but at the same time causing particular attraction between people characteristics. For example, it was possible to find out that vegetarians form a stable pair rather than lovers of hamburgers, and the area of ​​the face in a photograph in a certain way affects mutual attractiveness.
The company also analyzes the effectiveness of marketing companies' channels, uses personalized advertising, manages user loyalty, and counteracts the outflow of customers.
The company uses SPSS solutions, recommendation systems, Hadoop, Hive, and cloud technologies in data analysis.
Result: Every day, the system makes about 100 million assumptions that two people can fit together, $ 10 million is saved annually by countering customer outflows and by reducing inefficient marketing costs.
Branch: chemical industry, production of paints and varnishes.
Nippon Paint is a Japanese company, ranked 7th in terms of turnover in the world among paint manufacturing companies, the leading company in the Chinese market. Using the website launched by the company iColor , which allows you to try out different colors of paint on real interiors, has gained wide popularity among both private clients and designers and design companies. Analysis of the data obtained through this site allows the company to track new trends in colors and design in order to develop new products and plan production. Also, using this platform, the company interacts with designers and design companies to promote company products through them, segments consumers and creates personalized offers for them. For data processing and analysis, the company uses solutions based on SAP HANA and Hadoop.
Result: the company received a powerful tool for identifying and tracking market trends, allowing it to plan demand and develop new products, as well as a platform for solving a number of other tasks of targeted interaction with customers.
Industry: Health.
In the US, sepsis is ranked 10th in the ranking of causes of death among diseases. Each year, sepsis occurs in approximately 1 million Americans, from 28% to 50% of them die. The treatment of sepsis annually spends about $ 20 billion.
At the same time, the main cause of deaths is insufficient medical control. Patients are discharged from the hospital or receive first aid, and after that they are not observed. However, after this, there is a high risk of developing sepsis, the symptoms of which — fever, chills, rapid breathing and pulse, rash, confusion and disorientation — are similar to the symptoms of other common diseases. Often patients go to a doctor too late or the disease cannot be correctly diagnosed in the early stages. As a result, septic shock develops rapidly and often irreversible damage occurs to many organs.
To monitor the patient's condition, it is proposed to use a certified HealthPatch device from Vital Connect, which will collect key indicators of the patient's condition, including even postures and movements (they change in sepsis). The information is then sent to ClearStory Data servers, where it is combined with other medical patient data and analyzed in real time using an Apache Spark solution. In the future, such devices will be received by all patients who leave hospitals and receive first aid on conditions that may be followed by sepsis. A similar system, but with a lower level of data analysis, has already been successfully implemented in Singapore.
Result: a solution was created that will allow the US healthcare system to significantly reduce mortality from sepsis (general blood infection).
Industry: food delivery b2b.
JJ Food Service is one of the largest British b2b food delivery companies, having more than 60 thousand customers in the form of cafes, restaurants, school and office canteens, etc. In 2010, the company took almost all orders through call-centers, today 60% of orders are received via the Internet portal. This increased the efficiency of work, however, led to the loss of personal contact with the client. By phone, the client was offered to purchase more expensive or complementary goods and services, informed him about trends in his market segment. These capabilities needed to be implemented at a new level with the help of big data technologies. To solve these problems, the company turned to Microsoft to build a solution based on Azure Machine Learning cloud services. The recommendations formed by the predictive models built on this platform are now used not only on the Internet portal, but also by the call center employees. When a customer calls the call center or enters the site, his basket is already full based on the purchase history and recommendations (recipes, similar orders from other users are taken into account, new items are added that customers do not yet know about). About 80% of these goods are actually left in the basket and purchased by customers. Such a solution is possible in the b2b industry, because often enough regular supplies of a specific set of products are needed. Immediately before placing an order, the type of establishment and the recipes used by it are taken into account in order to determine whether the customer has forgotten to buy something necessary. The implementation of the system took 3 months.
Result: 80% of goods in pre-filled consumer baskets are purchased by customers, sales growth, speed of service and customer satisfaction.
Industry: banking.
In the previous post we have already described some cases of using big data in Sberbank, in this one we will tell about another case - AS SAFI.
This photo analysis system for customer identification and document fraud prevention was developed and implemented in Sberbank by early 2014. The system is based on comparing photographs from the database with images obtained by webcams on racks using computer vision technology. As a result, losses from this type of fraud have decreased by 10 times.
The basis of the AU was the biometric platform "Cascade Search" from the company "Technoserv". Initially, this system was designed for use in operational, reference and expert work, but was adapted for the needs of Sberbank and integrated with an automated credit application review system. The system works very quickly: thanks to a number of innovative solutions, such as In-Memory Processing, the comparison of camera images and images in the database takes only a few seconds.
Result: the loss from fraud with documents of individuals decreased by 10 times.
Branch: agriculture.
FarmLogs is a company that provides big data farmers with analytics and convenient services for planning and optimizing their work. Mobile and web applications of the company are already used by more than a third of US farmers. FarmLogs services use open geodata about soil type, detailed data on weather conditions, precipitation and solar activity. The analysis of satellite images is also widely used to automatically determine the crops of various crops, monitor their condition and take historical data into account for forecasting and making recommendations. The company provides farmers with rental devices that are installed in agricultural combine harvesters and automatically records in the system data on all the operations, routes and fuel consumption. As a result, detailed recommendations on agriculture are formed and all necessary calculations, including financial ones, are automatically carried out.
The result: a relatively far from high-tech farmer sector of the US economy is covered by big data optimization by more than a third.
Branch : transport - airports.
Dubai International Airport is the world's busiest airport in terms of international passenger traffic, one of the largest in the world. Large data on airport performance, flights and passenger movements are widely used to optimize airport operations and increase passenger satisfaction.
The airport uses sophisticated optimization algorithms to dynamically assign exits for embarkation and arrival. In particular, if two flights have a large number of passengers transfering from one flight to another, their exits will be assigned nearby.
Many passengers visit Dubai for shopping, and the airport has many shops in the duty free zone. However, the passion for shopping leads to the fact that passengers are often late for flights. Many of them do not speak either English or Arabic — the languages ​​in which the announcements are made. After the implementation of the new alert program in each of the airport’s shops, the boarding passes of the passengers are scanned, and they receive alerts about which exit they need to go, which route, and how long it will take, in the language they speak.
Result: the designation of departure gates and arrival was optimized, the number of delays on flights was significantly reduced.
Industry: retail, clothing, shoes and accessories.
Macy's is a large network of department stores, founded in 1858 and today has more than 840 stores in 45 US states. During the year, at least 1 time 70% of Americans visit the network stores.
The company analyzes large amounts of data on demand, stocks, lack of goods in specific stores, combines these data with the preferences of customers living in the area, and thus optimizes the range of all categories of goods in each of the outlets.
Using technology SAS Institute, the retailer makes the correction of prices for 73 million products almost in real time, using data on demand and inventory available.
The company uses big data not only in terms of internal optimization, but also in terms of customer orientation. For example, Macys.com online store, like other online retailers, uses personalization, targeted advertising banners and emailing, website search engine optimization, so that the buyer can easily find the product he is interested in. Personalization of advertising messages and offers in the company is very high, the number of unique variations of one mailing list can reach 500,000.
Result: high growth rates of sales (up to 50% per year), at least 10% of which, according to the company, is the net effect of big data.
Industry: data bank, help system.
The main asset of Ancestry is a huge database of modern and historical documents, allowing to restore family ties between people and build family trees. Today, the company's data bank already contains more than 5 billion profiles of people who lived at different times (from the 16th century) in a large number of countries, more than 45 million genealogical trees were built, specifying family ties between them.
Historical data is usually not in machine-readable format. This may be, for example, handwritten entries in account books. In addition, such data may be inconsistent, inaccurate and incomplete. The data mining and machine learning algorithms, including fuzzy matching algorithms, help the company in processing, supplementing and verifying data.
Big data technology is used to store and analyze data. Ancestry data is processed on three MapR clusters (a distribution based on Hadoop). The first one compares with the samples in the database the results of DNA analysis (there are already more than 120 thousand samples in the database), which users can receive for as little as $ 99, spitting into a tube and sending a sample of saliva to the company by mail. The second implements machine learning algorithms, the third - data mining.
Preliminary base analysis allows you to simplify the search for relatives and formulate assumptions that facilitate the investigation of the family history. Today, a huge number of discoveries made by users are provided by pre-combining profiles and ranking the results made by the system based on the analysis of profile data and the previous search history in the system. A few years ago, all such discoveries were the merit of only users of the system. The user experience in the system is also constantly analyzed to identify stages where users have difficulty to add additional content to these areas or rank search results.
Result: a huge database of historical data has been accumulated with cleaned, validated and pre-connected records, ensuring easy and productive use of the system.
Industry: mechanical engineering, engine manufacturing.
Rolls-Royce Holdings is a British multinational company producing engines and turbines for the aerospace industry, ships and energy. The engines that they produce are very large and expensive, miscalculations and mistakes in their production can cost millions and lead to death. Rolls-Royce uses big data technology to design, manufacture and further support its engines after sales.
In the development of engines is widely used computer simulation, producing terabytes of data. Analysis and visualization of this data is performed on high-performance computing clusters. The company's production systems interact with each other through the Internet of Things.
Rolls-Royce engines are equipped with hundreds of sensors that capture the smallest details of their work. This data is promptly processed using machine learning algorithms and transmitted to engineers in case of deviations.
The company does not provide an accurate assessment of the effectiveness of the implementation of big data technologies, however, according to statements by the management of the company, they gave a significant reduction in costs. Big data also changed the company's business model: thanks to them, the company was able to offer customers a new model of “Total Care” casing, when companies pay for hourly monitoring of engine performance during their operation.
The result: a significant reduction in development and production costs, increased reliability, the introduction of a new business model.
We welcome your comments.
And we are waiting for you on the Big Data Specialist program, starting on March 15.
Customer orientation
1. Company: Beeline.
Industry: Telecom.
Beeline has a huge set of data about its subscribers, which the company plans to use not only for internal optimization and work with clients (sales increase, customer retention, anti-fraud), but also to bring new analytical products to the market (providing data for credit scoring, targeted digital advertising, creating geo-analytical reports, IPTV analytics, external consulting). The company has implemented many big data projects. For example, the subscriber base was segmented on the basis of an extended client profile, including with the prediction of gender and age category, the construction of social graphs; a project was implemented to recognize and protect subscribers from money frauds and viral activities; subscribers who use communication services on several types of devices, as well as subscribers located at the airport and flying abroad, were offered to offer them suitable services and tariff plans. And this is not a complete list. The company uses HDFS and Apache Spark for data storage and processing, rapidminer and python for data analysis, including the scikit-learn library.
Result: by 2018, the projected revenue from big data will be more than 20% of the company's revenue.
')
2. Company: Caesars Entertainment.
Industry: entertainment, gambling.
Caesars Entertainment is one of the leaders in the US gambling business. It operates the famous casino Caesar's Palace in Las Vegas and over 50 gambling establishments around the world. The company itself has recently been fined more than $ 20 million for money laundering, and its casino division has been subject to bankruptcy. However, for 17 years, the company has accumulated an impressive amount of data and analytics for its Total Rewards loyalty program, which is now one of its most valuable assets and was valued at $ 1 billion. In the gaming industry, data on each client is important: some players of the highest, seventh category according to the company's classification spent about $ 500,000 in casinos annually, and one businessman left $ 200 million in Las Vegas for the year. It is very important to be able to meet such a customer by name, hold on your favorite game and offer free additional options to his taste: entertainment, air tickets, limousine trips, hotel accommodation, etc. With the help of accumulated data on the tastes and behavior of customers and advanced methods of data analysis, a company could find such an approach to customers that they leave as much money as possible and come back again. The company uses Hadoop and cloud solutions, processing more than 3 million records per hour. Data analysis is also used to segment customers and improve security standards.
Result: the most valuable individual asset of the company worth more than $ 1 billion has been created, the growth of profitability and safety standards has been achieved.
3. Company: eHarmony.
Industry: dating sites.
eHarmony is a dating site focused on building long-term relationships. In the form on the site, the user can leave a very detailed information about yourself, specify more than 1000 parameters. Further, using these data, as well as the history of sympathies and the development of relations between the users of the site, the system recommends the most suitable dating options, and not just on the basis of similarity of interests and beliefs. User photos are analyzed, complementing each other and coinciding, but at the same time causing particular attraction between people characteristics. For example, it was possible to find out that vegetarians form a stable pair rather than lovers of hamburgers, and the area of ​​the face in a photograph in a certain way affects mutual attractiveness.
The company also analyzes the effectiveness of marketing companies' channels, uses personalized advertising, manages user loyalty, and counteracts the outflow of customers.
The company uses SPSS solutions, recommendation systems, Hadoop, Hive, and cloud technologies in data analysis.
Result: Every day, the system makes about 100 million assumptions that two people can fit together, $ 10 million is saved annually by countering customer outflows and by reducing inefficient marketing costs.
4. Company: Nippon Paint.
Branch: chemical industry, production of paints and varnishes.
Nippon Paint is a Japanese company, ranked 7th in terms of turnover in the world among paint manufacturing companies, the leading company in the Chinese market. Using the website launched by the company iColor , which allows you to try out different colors of paint on real interiors, has gained wide popularity among both private clients and designers and design companies. Analysis of the data obtained through this site allows the company to track new trends in colors and design in order to develop new products and plan production. Also, using this platform, the company interacts with designers and design companies to promote company products through them, segments consumers and creates personalized offers for them. For data processing and analysis, the company uses solutions based on SAP HANA and Hadoop.
Result: the company received a powerful tool for identifying and tracking market trends, allowing it to plan demand and develop new products, as well as a platform for solving a number of other tasks of targeted interaction with customers.
5. Company: Hitachi Consulting, Vital Connect, ClearStory Data.
Industry: Health.
In the US, sepsis is ranked 10th in the ranking of causes of death among diseases. Each year, sepsis occurs in approximately 1 million Americans, from 28% to 50% of them die. The treatment of sepsis annually spends about $ 20 billion.
At the same time, the main cause of deaths is insufficient medical control. Patients are discharged from the hospital or receive first aid, and after that they are not observed. However, after this, there is a high risk of developing sepsis, the symptoms of which — fever, chills, rapid breathing and pulse, rash, confusion and disorientation — are similar to the symptoms of other common diseases. Often patients go to a doctor too late or the disease cannot be correctly diagnosed in the early stages. As a result, septic shock develops rapidly and often irreversible damage occurs to many organs.
To monitor the patient's condition, it is proposed to use a certified HealthPatch device from Vital Connect, which will collect key indicators of the patient's condition, including even postures and movements (they change in sepsis). The information is then sent to ClearStory Data servers, where it is combined with other medical patient data and analyzed in real time using an Apache Spark solution. In the future, such devices will be received by all patients who leave hospitals and receive first aid on conditions that may be followed by sepsis. A similar system, but with a lower level of data analysis, has already been successfully implemented in Singapore.
Result: a solution was created that will allow the US healthcare system to significantly reduce mortality from sepsis (general blood infection).
6. Company: JJ Food Service.
Industry: food delivery b2b.
JJ Food Service is one of the largest British b2b food delivery companies, having more than 60 thousand customers in the form of cafes, restaurants, school and office canteens, etc. In 2010, the company took almost all orders through call-centers, today 60% of orders are received via the Internet portal. This increased the efficiency of work, however, led to the loss of personal contact with the client. By phone, the client was offered to purchase more expensive or complementary goods and services, informed him about trends in his market segment. These capabilities needed to be implemented at a new level with the help of big data technologies. To solve these problems, the company turned to Microsoft to build a solution based on Azure Machine Learning cloud services. The recommendations formed by the predictive models built on this platform are now used not only on the Internet portal, but also by the call center employees. When a customer calls the call center or enters the site, his basket is already full based on the purchase history and recommendations (recipes, similar orders from other users are taken into account, new items are added that customers do not yet know about). About 80% of these goods are actually left in the basket and purchased by customers. Such a solution is possible in the b2b industry, because often enough regular supplies of a specific set of products are needed. Immediately before placing an order, the type of establishment and the recipes used by it are taken into account in order to determine whether the customer has forgotten to buy something necessary. The implementation of the system took 3 months.
Result: 80% of goods in pre-filled consumer baskets are purchased by customers, sales growth, speed of service and customer satisfaction.
Internal optimization
1. Company: Sberbank.
Industry: banking.
In the previous post we have already described some cases of using big data in Sberbank, in this one we will tell about another case - AS SAFI.
This photo analysis system for customer identification and document fraud prevention was developed and implemented in Sberbank by early 2014. The system is based on comparing photographs from the database with images obtained by webcams on racks using computer vision technology. As a result, losses from this type of fraud have decreased by 10 times.
The basis of the AU was the biometric platform "Cascade Search" from the company "Technoserv". Initially, this system was designed for use in operational, reference and expert work, but was adapted for the needs of Sberbank and integrated with an automated credit application review system. The system works very quickly: thanks to a number of innovative solutions, such as In-Memory Processing, the comparison of camera images and images in the database takes only a few seconds.
Result: the loss from fraud with documents of individuals decreased by 10 times.
2. Company: FarmLogs.
Branch: agriculture.
FarmLogs is a company that provides big data farmers with analytics and convenient services for planning and optimizing their work. Mobile and web applications of the company are already used by more than a third of US farmers. FarmLogs services use open geodata about soil type, detailed data on weather conditions, precipitation and solar activity. The analysis of satellite images is also widely used to automatically determine the crops of various crops, monitor their condition and take historical data into account for forecasting and making recommendations. The company provides farmers with rental devices that are installed in agricultural combine harvesters and automatically records in the system data on all the operations, routes and fuel consumption. As a result, detailed recommendations on agriculture are formed and all necessary calculations, including financial ones, are automatically carried out.
The result: a relatively far from high-tech farmer sector of the US economy is covered by big data optimization by more than a third.
3. Company: Dubai Airports.
Branch : transport - airports.
Dubai International Airport is the world's busiest airport in terms of international passenger traffic, one of the largest in the world. Large data on airport performance, flights and passenger movements are widely used to optimize airport operations and increase passenger satisfaction.
The airport uses sophisticated optimization algorithms to dynamically assign exits for embarkation and arrival. In particular, if two flights have a large number of passengers transfering from one flight to another, their exits will be assigned nearby.
Many passengers visit Dubai for shopping, and the airport has many shops in the duty free zone. However, the passion for shopping leads to the fact that passengers are often late for flights. Many of them do not speak either English or Arabic — the languages ​​in which the announcements are made. After the implementation of the new alert program in each of the airport’s shops, the boarding passes of the passengers are scanned, and they receive alerts about which exit they need to go, which route, and how long it will take, in the language they speak.
Result: the designation of departure gates and arrival was optimized, the number of delays on flights was significantly reduced.
4. Company: Macy's.
Industry: retail, clothing, shoes and accessories.
Macy's is a large network of department stores, founded in 1858 and today has more than 840 stores in 45 US states. During the year, at least 1 time 70% of Americans visit the network stores.
The company analyzes large amounts of data on demand, stocks, lack of goods in specific stores, combines these data with the preferences of customers living in the area, and thus optimizes the range of all categories of goods in each of the outlets.
Using technology SAS Institute, the retailer makes the correction of prices for 73 million products almost in real time, using data on demand and inventory available.
The company uses big data not only in terms of internal optimization, but also in terms of customer orientation. For example, Macys.com online store, like other online retailers, uses personalization, targeted advertising banners and emailing, website search engine optimization, so that the buyer can easily find the product he is interested in. Personalization of advertising messages and offers in the company is very high, the number of unique variations of one mailing list can reach 500,000.
Result: high growth rates of sales (up to 50% per year), at least 10% of which, according to the company, is the net effect of big data.
5. Company: Ancestry.
Industry: data bank, help system.
The main asset of Ancestry is a huge database of modern and historical documents, allowing to restore family ties between people and build family trees. Today, the company's data bank already contains more than 5 billion profiles of people who lived at different times (from the 16th century) in a large number of countries, more than 45 million genealogical trees were built, specifying family ties between them.
Historical data is usually not in machine-readable format. This may be, for example, handwritten entries in account books. In addition, such data may be inconsistent, inaccurate and incomplete. The data mining and machine learning algorithms, including fuzzy matching algorithms, help the company in processing, supplementing and verifying data.
Big data technology is used to store and analyze data. Ancestry data is processed on three MapR clusters (a distribution based on Hadoop). The first one compares with the samples in the database the results of DNA analysis (there are already more than 120 thousand samples in the database), which users can receive for as little as $ 99, spitting into a tube and sending a sample of saliva to the company by mail. The second implements machine learning algorithms, the third - data mining.
Preliminary base analysis allows you to simplify the search for relatives and formulate assumptions that facilitate the investigation of the family history. Today, a huge number of discoveries made by users are provided by pre-combining profiles and ranking the results made by the system based on the analysis of profile data and the previous search history in the system. A few years ago, all such discoveries were the merit of only users of the system. The user experience in the system is also constantly analyzed to identify stages where users have difficulty to add additional content to these areas or rank search results.
Result: a huge database of historical data has been accumulated with cleaned, validated and pre-connected records, ensuring easy and productive use of the system.
6. Company: Rolls-Royce Holdings.
Industry: mechanical engineering, engine manufacturing.
Rolls-Royce Holdings is a British multinational company producing engines and turbines for the aerospace industry, ships and energy. The engines that they produce are very large and expensive, miscalculations and mistakes in their production can cost millions and lead to death. Rolls-Royce uses big data technology to design, manufacture and further support its engines after sales.
In the development of engines is widely used computer simulation, producing terabytes of data. Analysis and visualization of this data is performed on high-performance computing clusters. The company's production systems interact with each other through the Internet of Things.
Rolls-Royce engines are equipped with hundreds of sensors that capture the smallest details of their work. This data is promptly processed using machine learning algorithms and transmitted to engineers in case of deviations.
The company does not provide an accurate assessment of the effectiveness of the implementation of big data technologies, however, according to statements by the management of the company, they gave a significant reduction in costs. Big data also changed the company's business model: thanks to them, the company was able to offer customers a new model of “Total Care” casing, when companies pay for hourly monitoring of engine performance during their operation.
The result: a significant reduction in development and production costs, increased reliability, the introduction of a new business model.
We welcome your comments.
And we are waiting for you on the Big Data Specialist program, starting on March 15.
Source: https://habr.com/ru/post/318208/
All Articles