Why do we need even more data centers: today and tomorrow big data analysts

Why store so much data in more and more data centers under construction? One of the applications of big date is predictive analytics. She answers the questions: what do these numbers mean about us, where is the analyst now used and what will happen in three years?

Forecasting - the basis of optimization

The amount of data grows at a rate that a person cannot imagine. Data is nothing without analysis. Only an unimaginable amount of information encoded in units and zeros. Why build new data centers? What and why is stored and also handles in their depths?

We all have heard about contextual advertising, the display of which is based on our preferences, which search engines will learn from our online activities. But about the rest of the sphere, few speak to the general public. But apart from the fact that a big date in the amount of predictive analytics allows advertisers and banks to earn incredible money, they help save human lives.
')

Where do without intrusive advertising?

We are still a commercial company, we don’t have sponsors and at the beginning of the article we would like to make to the readers an offer that is hard to refuse.

So, our hot offers, which in our opinion are interesting for the price, are simple configurations:

• Intel Core i7;
• frequency up to 3.8 GHz in turbo mode;
• 8 gb DDR3 RAM;
• 240 SSD;
• 1 gb / s unlimited for just 3.5 tons. Rub per month.

There are stones with the K index, which have an unlocked multiplier and overclocking is possible, but we do not approve of this, of course, and since you will have a constant and round-the-clock IP KVM, we may not notice :)

→ Go to order

Server sale

I also wanted to bring to your attention several “real” server configs, for example, now, as we consider, “tasty prices” for:

• 2xXeon E5-2670 8 Core 16 streams with 16gb RAM and 2 to 1TB for just 11790

→ Order

Also for all Habr's readers and writers, we offer a 50% discount for the first month and 15% for the renewal of services for all VPS services and dedicated servers. You can get a discount using the HABRAHABR code.

Predictive analytics applications

Human Resource Management

In this area, you can predict the professional performance of a profile in a public network, interesting vacancies for job seekers. LinkedIn, CareerBuilder, Wikipedia use this.

The company Hewlett-Packard predict the likelihood of employee departure. They use predictive models to evaluate each employee around the world. And there are more than 330,000 people. Thanks to the data obtained, managers can take appropriate measures to ensure that an employee remains or, where this is not possible, take into account the care of a team member when planning work processes. Savings of the company is at least 300 million dollars.



The special forces of the USA also uses the big date analysis. They need to know which of the candidates will be suitable for highly specialized responsible work better than others - they invest money in the training of those selected. Two key factors are the ability to be wrung out more than 80 times and the hardness of character, which speaks more than IQ.

Health care

Here, large data stored in reliable data centers are processed in order to predict serious and not very illnesses, the speed of disease development, the effectiveness of drugs and ... even death.

FICO predicts which patient will follow the medication schedule. According to statistics, about 125,000 people die prematurely every year due to failure to comply with the treatment regimen. Significant figure, right?

Pfizer specialists analyzed huge data layers to develop a short and accurate test for self-diagnosis of erectile dysfunction.

The University of Stanford invented a prognostic system that detects breast cancer more accurately than doctors.

Google Flu Trends, using user search queries, can predict a wave of influenza at least a week earlier than the USCRP.



Prediction of death is needed to offer special advice in insurance companies or to decide on surgical operations.

Human language processing, thought recognition, psychology

Language is one of the most difficult materials for machine learning. And although with the advent of the Turing test, it has been relatively many years, computers still have a hard time understanding human humor. Big date analysis helps with automatic moderation of comments, blog insults, negative feedback about services. For example, PayPal foresees customer care, and Citibank automatically sorts problem reports. In FordMotors, the system determines the level of driver attention, and the US Air Force is working on a similar system based on infrared shooting.

Success in this direction have made some scientists. For example, a program has been developed at the Hebrew University in Jerusalem that will determine sarcasm in consumer reviews on Amazon with an accuracy of 83%. They learned to recognize lies at the University of Buffalo. According to their method, the system monitors the movement of the eyes of the speaker and judges the veracity with an accuracy of 82%. There are positive results even in the field of writing analysis. There are ways that allow to establish the truth or falsity of statements from persons related to the criminal investigations of the military police.



In fact, machines are capable of reading thoughts: they decode the fMRI information of the brain and determine which object you are thinking with tolerable accuracy above 80%.

Fight against fraud and crime

Companies that use forecast models avoid losing large sums of money. They reveal fraud attempts in tax returns, checks, bills and contracts, car insurance payments and false guarantee cases. This is used in tax offices, postal services, banks.

Due to the correct analysis of data in Maryland they know which of the convicts will commit a murder or become a victim, in the USA they reinforce patrols of areas with an expected surge of crime, decide who can be released early on the basis of the probability of a relapse.



Amazon provides employee access levels, and Chicago predicts whether a crime will be solved by the characteristics of the victim and the murder itself. This saves energy, time and money.

Marketing, Advertising and Internet

Perhaps this is the most open area where a big date is used about all of us and our preferences. Most of all, big dates are needed for targeted marketing and sales prediction. It is necessary for everyone, while those who are available use it Among these companies are banks, retail chains, casinos, and telecommunications companies. So, Cox Communications tripled the number of responses to their mailing lists. Predicting the demand for specific home television, internet, or telephony services guarantees a 50% income per year for the sum of investments in this activity.

Predictions also successes the planned sales of such companies as IBM, Hewlett-Packard, Sun Microsystems. Amazon sells more than a third of all products thanks to personalized recommendations. Music and film companies also use machine learning - so they know in advance which movies and songs are doomed to popularity.

There are also “less commercial” points of application of predictive analytics. Google reduces the number of spam messages and filter malfunction.

State activity, politics, education, non-profit sector. Do you know what presidential competitors predict during election campaigns? The possibility of influencing the voters, namely - their opinion. The analyzed information helped Obama in 2012 to understand which of the voters would get a call, a meeting, a leaflet, or none of the things suggested. Individual approach gave their results.

Detection of safety and efficiency issues

Very important area. Here you can predict the probability of failure of complex systems of satellites, nuclear reactors, power supply, railway tracks, train wheels, building strength. And if the office equipment sales companies supply the warehouses with the necessary spare parts for the future, Con Edison, for example, takes into account accidents on sections of the energy distribution network, reducing risks.

Insurance and financial risk assessment

This is another one of the most profitable applications for big data analysis. The world's largest banks (Chase, Citigroup, Canadian Tire, PREMIER Bankcard) use forecast models to find out which of their borrowers can repay mortgage loans ahead of schedule, as well as to assess the solvency of potential borrowers. Algorithmic trading on stock exchanges is a legal way to make money using artificial intelligence.

Insurance companies (Accident Fund Insurance, Allstate) use prediction of injuries in accidents, costly injuries during work and even death.

Personal and family life

Divorce can be predicted with 90% accuracy based on clinical trial results. Also, university studies have found that infidelity in couples depends on behavioral characteristics and to some extent on genetic factors. Social networks use big dates to predict the likely friendship (Facebook and LinkedIn) and love (OkCupid predicts the likely polite response to messages depending on their content, and Match.com selects a pair based on data mining of user data).

Target chain stores can predict the pregnancy of women by analyzing their buying behavior. So they get 30% more potential clients - they offer them specialized advertising for parents.



Of course, this is not all cases of using big dates, their number is increasing every day.

Illustrative examples

Big date, machine learning, analytics - they are everywhere. This thought is reflected in the installation of Microsoft.

Microsoft / Infinity Room from Universal Everything on Vimeo .

Impressive, isn't it?

What will happen next? For example, after three years. His forecast was made by Eric Siegel, the founder of Predict Impact and the author of the book “calculate the future.” Here is how, in his opinion, the day of the person will look in the future:

On the way to the office, the only thing that predictive analytics is not involved in is how you turn the steering wheel with your hands, although this is only for now. When you sit on the seat, the system identifies you by reading biometrics. If a stranger got behind the wheel, the engine start would be blocked. Predictive system launches Internet radio with music that should suit your taste.



The autonavigation module establishes suitable routes, given the future congestion on the roads. On rises, maximum acceleration decreases automatically when the battery is low. Further, the computer system itself finds a convenient restaurant with the option of autodistribution, while it takes into account your taste preferences.

The secretary with artificial intelligence is responsible for the selection of news feeds that will interest you for sure. Invalid messages do not fall into your eyes, and are automatically filtered and do not waste time reading them. When you listen to voicemail, you do not receive commercial calls from robots, spam is also sorted.

Now the smartphone reads a letter from your operator, who wants to warn your early departure to a competitor. His offer is a big discount on the new iPhone 13. Today, the colleague’s son has a birthday and you give Siri an order to find a toy shop on the way. Siri has been improved and can be controlled through the car audio system.

If sensors “installed in your car” notice a deterioration in attention, the seat vibrates, telling you to focus on movement. If there is a chance of a collision, you will hear a loud sound and feel a stronger vibration of the chair. The car itself carries out diagnostics, predicts breakdowns and reports the need to get into the service center.

We hope that we have given a comprehensive answer to the question of why we need huge storage of big data.