In 2022, capture these 12 data and analyze trends!

author | Liu Yan

2022 year , Capture this 12 Data analysis trends ！

1 New perspectives on data and Analysis ： Build a new equation for business value

Many Chinese enterprises usually put 「 Data and Analysis 」 As a IT Noun . and Gartner Research findings of ,「 Data analysis 」 This topic has become more and more business level discussions . last year ,Gartner Put forward ： Data analysis capability is actually a business capability .

This year, ,Gartner A new slogan based on the whole data and analysis trend is proposed —— “ Build a new equation for business value ”.Gartner Express , By constructing such a new equation , It can help enterprises build more business capabilities based on data analysis .

This “ New equation ” How to understand ？

Gartner Sun Xin, senior research director, explained that , stay “=” To the right of , Is the growing demand for data analysis , Hope to create more value for enterprises through data analysis , Not just between one or two departments . We also hope that data analysis will become the origin of innovation , Bring new ideas to enterprises , Help enterprises move forward .

Gartner Find out , More and more realization based on data itself and cost reduction and efficiency increase have enabled enterprises to achieve revenue growth . The decision-making ability based on data analysis has become the core ability of a resilient enterprise .

Data analysis is not just about technology selection , Enterprises also want to use ““ The ability of data science to assist in making other decisions that fit the business context . Data analysis also brings a lot of thinking about business models , Help enterprises better carry out digital transformation .

Based on the above background ,Gartner hold 2022 The twelve trends of data analysis in are divided into three major themes .

Namely ：

One 、 Activate the vitality and diversity of enterprises .

Two 、 Enhance employees' ability and decision-making .

3、 ... and 、 Institutionalization of trust .

These three major themes contain four trends . Each trend does not exist alone , They are all linked up , And influence each other 、 Mutual reinforcement .

Theme 1 ： How to create a more dynamic 、 More resilient enterprises （ organization ）

Now? , Every enterprise has a large number of 、 Diverse data . However, enterprises often passively implement data analysis projects and behaviors , Not actively mining the potential value of data . Although it takes a lot of energy to store data , But these data have not been fully activated and utilized .

therefore ,Gartner Think , Enterprise data should be more dynamic 、 Use more diverse forms , So as to bring new opportunities and value to the enterprise .

Trend one ： Adaptive artificial intelligence system （Adaptive AI Systems）

from Gartner According to some research in the past few years , Many enterprises claim to have AI Action 、AI Ideas , But really put AI Few models operate . Recent years ,Gartner I've been talking about “AI engineering ” Action , Use engineering to make AI Gradually into the production environment , Thus operating value .

Gartner forecast ： To 2026 year , use AI Enterprises that build adaptive artificial intelligence systems by means of engineering , Compared with enterprises that have not done these measures , Operational AI There will be more models than 25%. This will undoubtedly bring more competitiveness to enterprises .

Gartner It is recommended that enterprises take advantage of something similar to Ops Means to establish AI Model , As is common Ops methods ,DevOps. Now? , More and more enterprises will use something similar to DataOps Data operation 、ModelOps Model operation , To build more effective AI Model .AI Engineering enabled adaptive AI System , It can help enterprises generate effective information more quickly AI Model .

Achieve more adaptive AI The system has a premise , This involves the second more important trend — Data centric artificial intelligence .

Trend two ： Data centric artificial intelligence

One of the existing challenges is , The AI solutions delivered by most enterprises largely depend on the availability or quality of data , And whether it can be understood by the business . Not just the artificial intelligence model , Many enterprises actually rely on Data Management .

However, many enterprises are establishing AI At the beginning of the model , There is no specific idea of what kind of data management model should be used to govern AI Model . In many AI projects , Data management is often underestimated . In some AI communities , Including the academic community and Industry , Many discussions focus on model development , While ignoring the tools and practices of data management . in fact , Data management capability can greatly improve the efficiency of the development and deployment of artificial intelligence .

Gartner Find out , Artificial intelligence with data as the core will continue to develop , It will expand more and more disciplines , Include ： Technology and skills of data management 、 Data quality 、 Data integration 、 Data governance ... These will expand into the basic capabilities of artificial intelligence . When one AI After the model is developed , The whole data management activity is not over 、 It will be continuously supported like a dynamic data pipeline AI Model development .

therefore , Enterprises need a more robust data management model to improve AI Ability to operate . I have to mention that , The next trend —— be based on “ Metadata ” Driven data weaving .

Trend three ： be based on “ Metadata ” Driven data weaving （Metadata-Driven Data Fabric）

“ Metadata ” What is it? ？ In short ,“ Metadata ” Is the data that depicts the data . How are these data used , What are the business implications of these data ？

In the past ,“ Metadata ” Be being “ Passive use ”, When enterprises encounter data quality problems , When it comes to governance data or governance requirements , Would be right “ Metadata ” Carry out a wider range of development .

Now? , Supplemented by some machine learning ability ,“ Metadata ” Can be highly utilized , Enterprises can make efficient use of metadata driven machine learning ability to do better data management , And then realize the situational data insight required by business development .

Besides , The problem of data islands has always existed , Enterprise data is scattered in various systems . This requires weaving the data , combining “ Metadata ” Driven machine learning , Plus the ability to use the knowledge map , Through such data discovery process, automatic data management can be realized , Reduce the workload of manual data management .

Now the data sources of enterprises are more and more abundant , Especially in “ cloud ” Data sources on the continue to generate , future , Enterprises will have to consider how to meet the needs of data management through more automated means . therefore , Next few years , We'll see more and more right “ Metadata ” Consideration , And use these “ Metadata ” The ability to complete better data discovery based on knowledge map .

Gartner Research findings , If we can use the means of data weaving 、 Manage data sources with metadata , It can effectively reduce many cumbersome data management work in the past . such as , Design of data pipeline , Data development , Data support , Quality of data, etc . meanwhile , To 2025 year , Data utilization will increase to 400%.

future , How much of the enterprise's data will be used by the business ？ This may not depend on the data itself , It's about metadata driven data management . With such a sound design pattern of data weaving , We can do better in data sharing . This is linked to the next trend ——“ Always share data ”.

Trend four ： Always share data （Always Share Data）

What do you mean “ Always share data ”？

In the past, a challenge faced by enterprises was , Worry about data risks when sharing data , Fear of data leakage . therefore , Many corporate executives reject sharing data , Dare not even share data .

In the research in recent years ,Gartner Find out , The concern of enterprises that data risk is greater than business value has almost dissipated . Many enterprises realize that , If you don't share data , The challenges faced by enterprises are surpassed by competitors , And the risk of digital execution failure becomes even greater . therefore , More and more enterprise executives are interested in data exchange 、 Data analysis capability 、 Focus on transforming data capability into business capability .

More and more enterprises are considering using data assets 、 Data directory 、 The data dictionary 、 Share data through data maps, etc , Share data in a way that can be governed . meanwhile , The direction of enterprise investment is also changing , Now enterprises will pay more attention to how to find more relevant data through automatic means , Can also use open OpenData Explore the possibility of self owned data more in the way of . The industry has also launched more open standards “ Metadata ”, In addition to sharing data , Also share how data can empower the business .

Topic 2 ： Pay attention to people , Enhance employees' ability and decision-making

The second big theme , Mainly from “ people ” Some thoughts from the perspective of .

It is people who ultimately make decisions in an enterprise , How to enhance the ability of these people , Help them improve their ability to make business decisions ？ This is the second major theme that enterprises need to complete .

In this topic , There is a big background that is irreversible , namely ： Enterprise data is becoming more and more complex , The environment in which enterprises use data is becoming more and more diverse . in other words , Users on the business side need more situational analysis , Become more and more urgent .

Trend five ： Scenario rich data analysis （Context Enriched Analysis）

Gartner forecast , To 2025 year , Context driven data analysis and artificial intelligence model , Will replace 60% Existing models based on traditional data . A trend reflected behind this is , More knowledge maps will be used . Now? , Knowledge map ” Already with more and more “ Metadata ” Management tools 、 Combination of prediction and analysis tools .

Through the map of knowledge , Can be carried out more accurately 、 More traceable prediction and Analysis , So that the data analysis of enterprises can be based on . Moreover, the embedded analysis of knowledge map can also bring more situations / Background information . The ability to map knowledge , More unstructured and unstructured data can be connected 、 Metadata , Help improve the situational conditions of data analysis .

Many enterprises are already using knowledge map to improve the ability of prediction and Analysis , Use knowledge atlas to collect situational information from more extensive data sources . The ability of situational analysis 、 Situational, richer analytical ability , It will become the analysis ability that enterprises must improve in the future .Gartner notice , Now many enterprises have begun to consider how to complete more data analysis through digital collaborative office software .

When it comes to data analysis, a lot of users on the business side begin to use data analysis , We have to mention the next trend —— From IT The transformation from embedded to business assembled data analysis .

Trend six ： from IT To business data analysis （From IT-Embedded to Business-Composed D&A）

Do data analysis on business , It's not a new topic for a long time . Let business users complete the data analysis of the whole life cycle , It has also begun to become possible . Because business users can now not only act as data analysts , Analyze your own data , Make some data insights ; Can also become a “ Low code 、 There is no code ” The application developers directly feed back the insight of data analysis to the business .

The past IT Embedded data analysis reports can gradually be done by employees on the business side , They can complete a data product or analytical application of data analysis by themselves .Gartner forecast , To 2025 year ,50% Embedded analytical content , It will be up to business users to take advantage of some low code 、 No code tool , Use assembled 、 Modular patchwork . The business side will be more business oriented to find data analysis capabilities , Finally, it becomes a data analysis application , No longer by IT（ The team ） To do everything .

The bottom-up in the past , Through the technology platform 、 The data warehouse built by the technology platform 、 database 、 Build semantic layer , Complete the work of the report , To a large extent, it will start thinking from the top-down business value flow of the business , Thus, the assembled data analysis ability of the enterprise becomes an analytical application .

From a human point of view , A big change is , In the past, report application was developed by application developers , Now there may be some “ Business technicians ” To complete ; More diagnostic types will be seen in the future 、 Predictive 、 Enhanced analytical content , Refined by business side users ; The pattern of writing code in the past , Maybe it's more command-line . Now use low code 、 Code free tools can be used in the simplest and easiest way , Build analytical applications ; The past technology may be solidified 、 The form of single software , In the future, more assembly technologies will be used to build applications .

The past may have relied more on SDK、API, Only developers can understand technology . future , More enterprises will be in the enterprise directory “ Building blocks ” As building blocks , Help business users build a business perspective ; From the perspective of design patterns , The past may have been more from IT See whether the report looks good from the perspective of 、 Whether it is easy to understand . In the future, business personnel will think more about their business based on their own , To operate and maintain data analysis products from the perspective of productization ; This can bring many more in line with business ideas 、 Situational data analysis application .

Trend seven ： Decision driven data analysis （Decision Centric D&A）

The seventh trend is , From data analysis driven decision making , Gradually transformed into decision driven data analysis .

In the past , Enterprises should first build the framework of data analysis , Then think about how to speed up the deployment of data analysis . Now? , Enterprises want to start with a business situation , Or the beginning of a business decision , Then think about how to use data analysis ability to help these decisions , Or say , Start from the last mile and imagine how to influence data analysis to business decisions .

This will involve a “ Integration team ” The concept of .“ Integration team ”, Refer to , Business and IT Can work together , Think about how to improve the path for enterprises to make data decisions .

Gartner A decision intelligence model has been proposed , Help enterprises manage the decision chain from the perspective of top-level design . Many enterprises may have made many reports , But I found that many people don't read these statements 、 Or many people think it's too rich , So hard to get to the point . therefore , Decision intelligence is for business and IT Integration team provides a way to improve organizational decision-making , Enterprises can use decision-making frameworks , Let the user analyze the data .

This trend , There is a prerequisite for implementation , namely , Enterprises need more and more people at a higher level , Make suggestions and plans based on data analysis for enterprise decision-making . This is related to the next trend ——“ Data and analysis skills 、 Lack of data literacy ” Related to .

Trend eight ： Data and analysis skills 、 Lack of data literacy （Data and Analytics Skills and Literacy Shortfall）

Gartner Find out , Enterprises generally have low data literacy , Probably IT（ The team ） Purchased a lot of tools 、 But the business is not really used .

Gartner forecast , To 2025 year , The chief data officer of most enterprises （CDO） Will not be able to develop in the workforce （ staff ） Have sufficient data literacy , To achieve a data-driven strategy . The lack of data analysis talents puzzles the managers of many enterprises .

stay Gartner Of “ Chief data officer research ” It is also found that , If an enterprise can take into account the factors of more people or personnel training , Will be better than not taking into account “ people ” The factors of enterprises are more likely to succeed . therefore ,“ people oriented ” Is the mission of data analysis , Enterprises need to cultivate （ staff ） Broader data literacy , Improve the ability of data analysis .

In view of the lack of data literacy ,Gartner Came up with a “ Three steps ” The plan , In the acquisition of talents 、 Cultivation of talents , Put forward countermeasures from three aspects of talent retention .

“ Access to talent ”： To motivate through business results , Let employees know that data analysis can help them solve more problems . Any data literacy training is not just to teach a variety of tools , The cultivation of tools is necessary . But more importantly , Let employees know , Using data analysis can help solve practical business problems , So as to improve work efficiency .

Train people ： Beyond seas , Cultivating good data analysis talents is mainly through community governance . Enterprises can establish a data analysis community , Let's discuss in the community , So that users and users can achieve “ To help ” The process of , Build a data culture through the community .

Retain talent ： At present, some enterprises have deployed human resources departments , Make it add some data analysis content to the daily work of employees , And adopt some incentives , Let employees use data analysis , Finally, the results of data analysis will be rewarded .

Theme 3 ： Institutionalization of trust

The third theme may be of great concern to Chinese enterprises .

We have discussed how to use data analysis on the business side . However, the premise that the business side can use data analysis is , They trust data , I believe that enterprises can give them such ability （ jurisdiction ）, And the data can be used directly 、 No responsibility . therefore , To achieve ubiquitous data analysis capabilities , It is very important to institutionalize trust .

The first trend under this theme is , Internet governance （Connected Governance）.

Trend nine ： Internet governance （Connected Governance）

“ Internet governance ”, It doesn't mean building a new team , Because in the enterprise , Is likely to “ Data governance 、 Security Governance ” Or the most common “IT government ”, Have established relevant capabilities .

“ Internet governance ” It's actually a framework , Used to build a cross organization 、 Cross business functions , Even cross regional 、 Virtual data and analysis governance layer , To achieve cross enterprise governance results .

In China, , A particularly obvious phenomenon is , With the establishment of local laws and regulations in China , The foreign governance model is completely inapplicable in the Chinese market , Or the governance challenge is even greater . Because we should not only consider foreign laws and regulations , Also take into account domestic laws and regulations .

Besides , There are more and more factors of governance , such as ： Data quality 、 Data security 、 Data privacy 、 Data ethics , For the definition model of data , The management of the whole life cycle is included in the scope of governance . therefore , For businesses , The way of Internet governance may be the measures that have to be taken .

Have “ Internet governance ” The pattern of , It is very important to build a broader data governance team . In many Chinese enterprises , Will consider establishing “ Chief data Officer ” Office , Set up a data governance committee under the office , The Department will work with the law in the future 、 Security cooperation . But the “ Internet governance ” At a higher level , Not scattered 、 Governed by governance treaties .

stay “ Governance elements ” in , The important thing is security and privacy . therefore ,Gartner Also put forward , about AI Trends in trust risk and security management .

Trend ten ：AI Trust risk and security management （AI Trust Risk and Security Management）

Gartner Previous research found that ,50% Of AI The model never entered the production environment . among ,“ Security ” and “ privacy ” Is one of the main reasons .

AI The innovation and speed of innovation are under a lot of internal and external pressure . For example, in order to maintain normal AI operation , Many enterprises cut corners in AI trust risk and security management , This can lead to some negative results . such as ： Will be fined ,AI Of ROI It will also be greatly reduced . therefore , Enterprises need to spend more time and resources on artificial intelligence risk and security management .

Most enterprises are developing AI Model time , Not clear about what you want to achieve . Many enterprises often do not have a complete process 、 Tools or metrics to govern and manage AI Trust and security risks .

in addition , Many enterprises tend to collect training data of artificial intelligence , But there is no reasonable goal in selecting data , At this time, the data often have some bias , These biases can have a negative impact on the quality of the data model .

Now many enterprises are driven by regulation and compliance , When doing model governance , Due to compliance control , The enterprise is doing AI The model is completely passive , This compliance does not necessarily lead to credible AI Model .

therefore ,Gartner Put forward this trend , It is hoped that enterprises will pay attention to trust risk and security management AI government .

Trend 11 ： Data and analysis ecology of manufacturers and regions （Vendor and Region Ecosystems）

Gartner In last year's survey, I saw , More and more enterprises are building their own localized or domestic data analysis capabilities . Based on this ,Gartner Put forward ,“ Data and analysis ecology of manufacturers and regions ” This trend .

The initial state for most enterprises to establish data analysis is , Choose to use a set of data analysis solutions all over the world . But with regional governance , Many enterprises need to be local / Establish a set of repeated in the region 、 Technology stack for data analysis in accordance with local terms . These technology stacks must meet some domestic requirements , This brings the challenge of model selection , On the one hand, we need to do data analysis and operation in combination with domestic ecology .

Gartner I also see another interesting trend , In the past , Many buyers 、 When Party A selects products , There is often a concern —— Whether you must choose a cloud vendor to build your own data analysis architecture . The main reason behind this is , I'm worried if I'm taken by a family “ Cloud manufacturers ” binding , There will be some business problems when renewing the contract in the future .

But now , More and more companies are finding , Choose a “ Cloud manufacturers ”, Use one “ Cloud manufacturers ” When analyzing ecological products , It involves data management 、 The trouble of analysis and management will be reduced a lot . therefore , Apart from the confusion of being bound by a manufacturer , More and more enterprises prefer to use one “ Cloud manufacturers ” The ecology of .

therefore , When enterprises establish data analysis Ecology , You can think more about , What abilities can be in one “ Cloud manufacturers ” Implement in . Or say , When some foreign enterprises enter China , When building a parallel data analysis stack , Whether to analyze the data in the cloud .

Now the capacity of ecological products for data analysis is growing , How to complete the construction of various capacities in a top-down way , Not just considering the construction of tools has become very important . So in the future , When enterprises establish their own data analysis Ecology , More consideration should be given to the compatibility between manufacturers . Domestic data analysis ecology will also become enterprise product selection 、 Important considerations in platform selection .

Trend 12 ： Data analysis in the edge （Data and Analytics Expansion to The Edge）

Data and analysis activities are more and more distributed equipment servers outside the data center or public cloud infrastructure 、 Operate in the gateway .

This year, , This trend is more obvious . A big reason for this is , Data analysis done in edge devices , More in line with the emphasis now “ Data sovereignty ” or “ regulatory ” The appeal of . therefore , I hope more and more data analysis , In terms of architecture, distributed architecture can be considered , To help complete more effective data analysis .

“ edge ” It's a continuum . from “ cloud 、 Data Center ” To “ Equipment edge ”, Not just a single location . analysis 、 Especially the use cases of artificial intelligence , It may be best placed in different positions of the continuum 、 place , Not in one “ spot ” On .

therefore , Managers of enterprise data analysis , May have to give up All in In the form of , Focus on deploying data analysis capabilities in some public clouds or data centers , Deploy in a distributed data analysis environment .

2 Talk about China's data and analysis trends

The epidemic has become an opportunity for the digital transformation of Chinese enterprises

Sun Xin thinks , The epidemic has become an opportunity for the digital transformation of many Chinese enterprises . In the past , Many enterprises have no strong demand for digitization , But in recent years , Digitization has become a rigid need . How to make more and more users work remotely , How to enable more and more users to make decisions based on data ？ Have become a thorny problem in front of enterprises .

Now , stay “ cloud ” It is a trend to do data analysis on . Extending to the world , stay “ cloud ” Data analysis on has become a default preference . Relatively speaking , The deployment of public cloud by Chinese enterprises is not so vigorous , But because of the epidemic , Domestic enterprises' upper limit data analysis ability of public cloud has increased significantly .

On the other hand . Business side users use data analysis to become a trend . Some domestic business intelligence BI Tool and data science tool manufacturers have achieved rapid growth this year . Enterprises are increasingly looking forward to taking advantage of “ Self service ” Tools for , Help business users make decisions faster , This is also the driving force for enterprises in the context of the epidemic .

The epidemic has accelerated the digitization process of many traditional enterprises , The resulting data processing 、 What are the requirements and challenges of data management different from before ？

Sun Xin is right InfoQ Express , In this respect , The demand for automation has increased , In the past, enterprises did not have such a high demand for automation in data management , But now , Enterprises have a growing demand for data , The timeliness of data reflected in the decision-making chain is becoming more and more urgent , It also relies on more automatic means to complete the automation of data management . After the outbreak , Enterprises have a higher demand for the speed of data management .“ Data weaving ” This method , The function of automation based on machine learning , Reduce manual workload , So as to speed up the speed of generating value from data .

Gartner We also see the urgent need of enterprises for analysis ability . In the past , Enterprises may just do some data visualization work , Now we may not be satisfied with just doing this kind of descriptive analysis . There will also be some diagnostic analysis 、 Exploratory analysis 、 Predictive analysis , These require more advanced analytical capabilities .

On the other hand , After the outbreak , Many enterprises urgently need , In the workflow 、 In digital office software 、CRM、ERP in , More quickly embed some data analysis applications , So that business users can make faster judgments . Business assembled 、 Modular analysis can help enterprises improve the ability of embedded analysis more quickly , So as to help enterprises run faster in business processes 、 Make better strategic judgments .

Big data and AI The integration of has become a hot trend

Talking about big data and AI The trend of integration , Sun Xin is right InfoQ Etc , Big data and AI This is closely related .

Big data and AI The combination of is embodied in several trends ：

Data Centric AI, Take advantage of better data management , drive AI Model development . Adaptive artificial intelligence system , Drive the development of high-quality products by absorbing environmental variables AI Model , Make it adaptive . Big data and AI Fusion , The closest point is , hold AI The ability of is embedded in the function of data analysis or big data , Complete more enhanced analytical capabilities .

“ Enhanced analysis ” It's the last few years Gartner Keep talking about trends .“ Enhanced analytics is not about letting users write AI Or data analysis code , It's about how to package it 、 Let users still use simple and easy-to-use forms 、 In the form of dragging 、 In the form of natural language , Do more advanced analysis . therefore , More and more people will see AI To empower big data products 、 industry , So that more people can do more in-depth analysis with a lower threshold , This is big data and AI Directly enable the performance of business side users .

Can edge computing be combined with big data platforms ？

On the current relationship between big data platform and edge computing , Sun Xinxiang InfoQ Response statement , Now there are many AI The ability of 、 The ability to analyze data , It needs to be on the hardware side . especially , Like some special chips, they can adapt very well AI Development of models and computing power . therefore , In the future, the effect of data analysis algorithm will appear more and more through the improvement of hardware capability , So as to help enterprises complete more efficient data analysis and development at the edge .

talk about ,2022 The trend of edge computing in , Sun Xin said , To 2025 year , exceed 50% Enterprise level core data , Will be created and analyzed outside the data center and cloud .” The main reasons are as follows 2 There are two contributing factors , One is , The need for automation and control of remote environment . In addition, there are security and resource constraints , And the increasingly complex demands of data sovereignty and regulation . Do data analysis at the edge , It will become an important trend .

The lake and the warehouse are integrated , To what extent has it developed in Chinese enterprises ？

How to combine the past data lake and data warehouse , It is a topic often discussed in the industry . however , There is a misunderstanding , Many enterprises just want to make their past unstructured 、 Structured data all exist together , Just build a lake warehouse integrated structure , But ignore the original intention of doing this thing .

What is the initial intention ？ Sun Xin said , Enterprises can go back , Why build a data Lake , Why can't shucang do this ？ because , in the past , The purpose of using data warehouse is to solve some known problems caused by known data and known models . Use data Lake , More to answer unknown data 、 Do not know how to model the data and some unknown problems to be predicted , The data will become more complex .

The lake and the warehouse are integrated , It brings more possibilities for enterprises to complete advanced analysis on an integrated platform , I hope more users can expand their analysis ability , In the past, the warehouse can only complete some descriptive analysis 、 Simple diagnostic analysis , Then it extends to some predictive analysis capabilities . therefore , Enterprises are now more concerned about , What is the output of Lake Warehouse Integration , How to enable users to complete more use cases of advanced analysis in the operation and maintenance environment .

therefore , In China, , Now many enterprises may blindly hear some “ The lake and the warehouse are integrated ” The concept of , In fact, you should calm down and think about it ：“ The lake and the warehouse are integrated ” What is the result ？