In fact, if companies can figure out some misunderstandings around big data, you may be able to help them avoid developing the wrong business development, and thus saved the day, to prevent waste a lot of time and money consuming market competitive position of, or damage to the enterprise reputation.
The following are some of the biggest myths around about big data understandable.
Myth 1: Only an expert can deal with large data scientific data
In fact, relying solely on expert scientific data itself is not enough.
"If they do not know their own from the beginning hopes big data analytics to find what, then, your business data employed by scientific experts to be able to obtain the kind of information from big data analysis is also helpless." University of Pennsylvania Medical data analysis, senior director Pat. Farrell said. "Enterprises need is those familiar with the industry knowledge of the field, what kind of memory to understand the industry in question, can get insight into how the value of a particular industry professionals from big data analysis."
For example, the University of Pennsylvania School of Medicine, including health and medical systems. Their health systems have spent a long time working on clinical data collected in the data warehouse. At the same time, its medical school using new technologies a lot of data research process sequencing of the human genome in need.
"We know that a large mass of data we collect certainly contains very valuable things, and we finally able to gain access to this valuable information by certain computing power." Farrell said. Combined with medical expertise for data analysis opens up a whole new field for their health care to predict, he said.
Myth 2: larger amount of data means greater value
Collect, store data, and its cataloging, it takes considerable time and resources, Farrell said. And if indiscriminately just a simple collection of large amounts of data information is often more valuable items will transfer.
Farrell suggested that companies prior to start collecting data, there must be a clear understanding of the specific targets or key performance indicators.
"Companies need to understand, it must be through the wisdom of massive data collected for analysis, and then dig out the value of the point." He said. "Just the data collection itself is not enough."
Myth 3: Big data is only to be useful only for large enterprises
Large companies may have more data sources from within the enterprise, but even small businesses can take advantage of government agencies, as well as data from social media platforms, data vendors.
"No matter your organization's size is how, when you make strategic decisions related to business development is based on the best data analysis, rather than merely relying solely on intuition." Dell Software Product Management, Information Management Solutions Executive Director Darin Bartik said.
Compared For larger enterprises, smaller companies may be less than their peers of large enterprises use data to make business decisions, but once small business to do so, they can make better and faster good business decisions.
"Small businesses can take advantage of best practice solutions become more data-driven enterprise, or outwit actually exceed those of larger, slow reaction data-driven competitors." He said.
Myth 4: It is now collected, after sorting
Now the store is really getting cheaper, but it is not free after all. And for many companies, they store data appetite to expand much faster than the speed of the storage cost reduction. San Francisco-based cloud BI vendors Birst CEO Brad Peters said.
Those who believe that just a simple data collection companies, the future will be how to deal with these massive amounts of data to worry about, he said. "I've seen some big companies to collect vast amounts of data information, its collection and storage costs continue to rise, but companies do not derive any valuable analytical insight."
In fact, for some data sets, it has begun to apply the law of diminishing returns. For example, you need to predict the election results based on the number of voters in certain elections, so you need voters Yidingshuliang to obtain a representative sample. But after you collect the number of voters exceeds a certain point, adding more people will not vote for a significant degree of influence election results to determine the error.
Brad Peters asked: "You alone company to collect more data, it may give you a more accurate predictive analysis of it?" Or collect more data means that you do not need businesses to hire more staff ? You can ensure that the network would be better for your business do? We collect data rate can not be too fast, can not exceed the enterprise economy and the budget increase speed. "
And this is not just a problem of storage costs, in San Francisco Recommind company specializing in unstructured data analysis of large data management and information management, global head of Dean Gonsowski said. For example, if the data is out of control, costs may make companies spend out of control, he said. In addition, the enterprise data warehouse to store more data means that companies need to take more responsibility to comply with data governance laws. Finally, the more enterprise data collection, the greater the need for these data collated. "When the database reaches billions of searches, the search time is bound to be extended, so that those who have never really go through a good deal of information will cause clogging of the system."
Myth 5: All data and have the same importance
In the past 20 years, Virginia has been collecting on student enrollment, degree-granting financial assistance and other relevant data. However, this does not mean that 20 years of data collected and stored in the same data field must be the same data.
"At present, the biggest problem I have to deal with that in the data dictionary, the researchers believe the importance of all data are equal." Policy Research Council, Virginia State Higher Education and data warehouse director Todd Massa representation . "For example, we ACT on students and SAT test scores data collection, initially only collect the students in the state information, and then we feel that this data gaps, began while collecting Honshu inside and outside the student's data." Similarly, we also collecting data to test the K-12 level students of different races and track their higher education situation.
In fact, for different institutions, or at different points in time, or for different people within these institutions, any particular data may have different importance. "If an isolated storage or companies need their data collected solely responsible, then you may have come across many different situations." He said. "However, the importance of data will change over time."
"As a result, analysts need to have more than just statistical skills, but also need to provide the data and trends in the industry as a whole to be analyzed, such as the SAT and ACT scores recalibration." He said.
You can not have all this data into a data warehouse. The same applies to external data sources, he added. "In the past 50 years, the data set at the federal level has undergone tremendous changes, understand the cultural background data acquisition is the premise of the need to use the data."
Myth 6: The more specific the better predict
Something that the more specific the more accurate is human nature. For example: "3:12 PM" to be more accurate than "sometime in the afternoon." Similarly, the forecast "Sunday morning will rain" than "this weekend 50 percent chance of rain" is more accurate.
In fact, the opposite is true. In many cases, a more accurate prediction is unlikely to be accurate.
Myth 7: Hadoop big data equals
Hadoop is a very popular open source database of unstructured data in recent've got a lot of attention. But in fact there are other business options.
"There is a whole NoSQL available for the enterprises." Irfan Khan Big Data SAP general manager and senior vice president, said. "At the same time, there is a whole rack of other technologies MongoDB, Cassandra and so on." Some of these techniques may be better suited for a particular project than the other big data technologies.
In particular, the working principle is to Hadoop data into blocks, and simultaneously on a plurality of data blocks. This method is suitable for many of the problems of big data, but not all the problems.
"Although YARN framework and the Hadoop 2 can resolve some of these issues, but sometimes you need a way to deal with the problem, Hadoop is not the best." A big data consulting company LucidWorks the CTO Grant Ingersoll said. "People need to remain calm, decide what technology is best for them, rather than simply rely on technology today is the most fashionable technologies to be judged."
Myth 8: End users do not need direct access to large data
As companies collect large amounts of data from a large variety of sources of high-speed, it seems that for full-time employees of enterprises, these big data have become quite complicated. but it is not the truth.
For example, in intensive care units, all of the data generated by the equipment, including heart rate, respiration data, ECG readings. Many times, doctors and nurses can only see the patient's current readings.
"I can not see 10 minutes before the reading is like, or draw a picture of the trend chart, understand from now on after an hour what the trend." Chief Marketing Officer of Philips Medical care and clinical information of patient care Antonio . Jones said. But to understand the historical patient information for physicians to make decisions related to medical programs is very valuable.
The question now is, we need to get all the data generated by different devices interact with each other, even if they were not originally designed for this. And even the use of different operating platforms, operating systems and programming languages. However, once you do, it can help doctors and nurses get more valuable data when needed.
9 misunderstanding: Big data is a big problem
Some of the major CIO large banks have recently started to discuss the topic of big data, and ask about end-user self-service.
Some executives believe that big data will only be able to answer certain types of questions. Their attitude can be summarized as: "Our goal is to solve the big data problem and a handful of high-value data through a set of core data scientists we do not want clutter, so that ordinary people are able to access and use this new information. I do not think the average person needs these data. "
Birst CEO Brad Peters does not agree with this approach, but he said it is common in many industries are in. "In many large insurance companies 'internal business users do not have many smart enough to handle big data' idea is rampant."
Myth 10: Big bubble will eventually burst data
With the hype cycle may come and go, but changes in technology will continue. The end of the dot-com bubble burst of the Internet is not a signal.
Even after the hype calmed down, companies will still require large data processing. In fact, as was the exponential growth of data, than they ever have a need for more large data processing. IDC estimates that by 2020, every two years the cumulative amount of data collected will be doubled.
Moreover, companies will not just collected more data than they are currently. At the same time, new data types can also occur at the same time, requires a lot of storage.
"We will have mapped the genome of the patient." Patient Care and Chief Marketing Officer of Philips Healthcare clinical information Anthony. Jones said. "It allows physicians to customize solutions according to the specific medical situation of the patient. When we talk about big data, we are referring not just to collect vast amounts of data. I do not think most of the CIO will really feel that the collection of data would be What a very difficult thing. "
Just a little hesitant about whether to adopt the "big data" projects, business is likely to miss the opportunity to capture the data elements will affect their business in, Bryan Hill Cadient Group CTO.
"Just as cloud computing, the term big data 'may change, that the Internet is no different, but the spirit of big data research will be forever." Bryan Hill said.
