Data governance in manufacturing is still in its early stages, and data quality management is an area of focus for all data-related projects. Data governance is based on the management of data standards, data quality and metadata, and is the basis for enterprises to realize the value creation of data assets. Last week, at the special training course on "Manufacturing Data Management and Application (Phase I)" of the "Ten Thousand Enterprises Cultivating Talents Project" sponsored by Wuhan Municipal Bureau of Economy and Information Technology, Yixin Huachen data governance expert Wen Laisong combined manufacturing Based on the status quo of digital transformation in the industry and his own work experience, he shared practical methods for improving the quality of manufacturing data.
01
The need to improve data quality
Data has become the driving force for the development of manufacturing enterprises, and high-quality data is the basis for supporting business management and business decision-making. The problem of data quality has been puzzling enterprise decision makers, managers and executives. In an enterprise, many "stories" occur around data every day, but there are also many "accidents".
According to statistical analysis, the causes of poor data quality include all links from data collection/entry to data processing/processing to data application presentation. Any error in any link will lead to data errors, and even the source data itself is wrong. Therefore, the problem of data quality is not only a technical management, it may appear in the process of business and management.
Causes of data problems in the whole process
As such, data quality improvement is particularly important in data management projects. Its purpose is to establish data standards for enterprises, improve and control data quality, and ensure the accuracy, consistency, integrity, uniqueness, timeliness, and effectiveness of enterprise data. To ensure the correctness of business data application and business decision-making.
In the context of "digital transformation of the manufacturing industry", facing data with rich sources and diverse characteristics, data fusion management is gradually becoming an important content and requirement for the construction and management of information channels for various departments of enterprises. Data quality improvement is a necessary step to improve leadership decision-making and governance capabilities, and to promote the operation of enterprise data resources.
02
The concept of data quality improvement
What is data quality?
DAMA's definition of data quality is: to ensure that the needs of data consumers are met, the application of data management techniques for planning, implementation and control and other management activities. DCMM regards data quality as a major capability domain, and plans secondary capability items including data quality requirements, data quality inspection, data quality analysis, and data quality improvement.
Data quality is the basis for ensuring data application. The industry has defined a data quality evaluation index framework, and the evaluation of data quality includes the following six dimensions:
Integrity: Refers to whether data information is missing.
Uniqueness: It means that any entity in the data set will not appear repeatedly.
Consistency: Refers to ensuring the degree to which data values are expressed within a data set and across data sets.
Accuracy: Refers to whether there is any abnormality or error in the information recorded in the data.
Timeliness: Refers to the update frequency of data in line with expectations.
Validity: refers to the degree to which the described data follows predetermined grammatical rules.
So what is data quality improvement?
Data quality improvement is usually understood as a series of data cleaning, conversion, and modification of business data in application systems and aggregated data in data warehouses, so that the data can be improved in integrity, uniqueness, timeliness, validity, accuracy, In terms of consistency, it is a process to meet data applications such as report query, decision support, early warning and forecasting. Data quality improvement is a process in which an enterprise, under the guidance of its data strategy, formulates its own data management system, reasonably plans the enterprise data structure, and conducts a series of standardization and data governance work through a professional platform to make data an enterprise asset.
Data quality improvement is part of data management, and data quality improvement includes three aspects of construction:
-
Management system construction: By optimizing the data governance organizational structure, formulating data governance system specifications, forming an enterprise data governance system, and promoting the effective improvement of data quality
-
Platform capacity building: provide efficient technical means for data quality improvement, break down data islands, and realize efficient management of enterprise data on the same platform
-
Data governance implementation: improve data quality, create data assets that continue to appreciate in value, and improve high-quality data services. Turn data into readable and easy-to-understand content for the whole group, and quickly integrate it into business, strengthen data application capabilities, and turn data into productivity
Stage division of data quality improvement
03
Ways to Improve Manufacturing Data Quality
The core of the digital transformation of the manufacturing industry lies in data, and data linking, aggregation, and governance are the first steps in the digital transformation of manufacturing enterprises. How can manufacturing companies improve data quality? After more than 10 years of exploration, Yixin Huachen, as a leader in the field of data governance, now has a set of effective methods for managing data quality. The specific implementation steps are as follows:
Standard process for data quality improvement
Step 1: Investigate data issues
Carry out questionnaires for relevant business departments and IT systems, collect obstacles and challenges in the use of information, and clarify the status of data and user needs. Based on the survey questionnaire, special face-to-face interviews are organized for the areas that the project initiators focus on, from the source system, data application system, business department, technology department to the data governance leader, gradually establish the overall architecture view, and outline the target blueprint.
Step 2: Data Governance Maturity Assessment
The stages of enterprise data maturity are divided into data accumulation stage, initial stage, system stage, quantitative management stage, and continuous optimization stage. It is very necessary for enterprises to recognize their own data maturity. At present, the informatization construction of the manufacturing industry has achieved remarkable results. Enterprises generally hope to lay a solid foundation for the construction of data applications through data governance, so as to realize the value of data.
Step 3: Identify governance objectives and obtain top-level support
Data governance is the number one project. Data governance is a cross-departmental, cross-system activity that requires significant support from the top.
Step 4: Develop a data governance implementation roadmap
Step 5: Build a data governance organizational system
Organize a cross-departmental data governance working group, determine organizational goals and positioning, determine organizational form and hierarchy, clarify management content and division of responsibilities, and set positions and personnel arrangements.
Step 6: Sort out the business system and find out the family background
Not only the data, but also a "comprehensive physical examination" of the business process is required to sort out the business system and find out the status quo of the data.
Step 7: Build a Metadata Repository
Step 8.1: Base Standard Definition Template
Step 8.2: Data quality management closed-loop mechanism
In the development process of quality management, many quality management theories have been formed, and the PDCA cycle carries out quality management through four repeated steps of planning, execution, inspection, and processing. PDCA is mapped to the closed-loop management of data problems, including data problem discovery, problem location, problem tracking, improvement of problem knowledge base, and assessment and evaluation, forming a sustainable operation and sustainable problem-solving mechanism.
Step 8.3: Master Data Management
Step 9: Establish a long-term mechanism for data governance
Data quality management runs through the entire life cycle of data. It is a long-term and continuous work that requires the joint efforts of business personnel and technical personnel to obtain high-quality data. Therefore, it is necessary to establish a long-term mechanism from the following aspects to continuously optimize and iteratively improve the value of data:
1. Promoting data capitalization, so that data strategy and business strategy can be integrated and unified
2. Visual display of data governance, intuitive presentation, analysis from various dimensions of the data governance domain
3. Evaluate the effectiveness of governance, measure the effectiveness through results, and report to the committee regularly
4. Promote data governance culture, publicize data strategies, and establish data forums
5. Establish a systematic data service system and form a unified data service entrance
Example of Governance Effectiveness Evaluation
04
Implementation of Data Quality Improvement
The manufacturing industry has begun to gradually explore data governance and management applications, and has achieved practical results. Yixin Huachen is actively deploying the field of data governance, using DCMM and DAMA as the theoretical framework, and combining practice to build a closed-loop data management system. Data governance products have been successfully applied to many enterprises.
Case 1: Data governance project of an OEM
Customer pain points: With the continuous advancement of car manufacturing technology, car companies have entered the era of customization, and they need to produce diversified products at a lower cost to meet the various needs of different customers. As far as the data itself is concerned, the growth of business volume has accelerated the speed of data expansion. Therefore, data governance has become the general trend.
Solution: Create a digital operation system at the level of the car enterprise group and carry out the overall design according to "119". Efficiently accumulate data assets, empower business application scenarios, and help enterprises build a solid data foundation and realize digital operations.
-
Consulting and planning: Combined with the theoretical organizational structure, establish a three-tier governance organizational structure for car companies' decision-making, management, and execution, including: governance management committee, governance management department, and executive working group. The output of consulting results is divided into two aspects: one is the data governance system, and the other is the data governance management method.
-
Platform construction: The data governance platform is built on the data base, collects metadata and master data models of various systems and services of car companies, builds standards, conducts quality control, realizes asset-based operations, and empowers business departments and application innovation.
-
Governance service: Based on the data governance business needs of car companies, through data collection, master data, metadata, data standards, data quality, data security, data asset management, data services, data indicators and other comprehensive governance, the value of data will be improved.
Governance results:
1. Construct data links between 6 major business areas and 11 core application systems, display dependencies between systems through data maps, understand the business meaning of data layer by layer, and understand the scope of influence through metadata analysis
2. Establish a data quality inspection & rectification mechanism, complete the data quality inspection work in the data entry and data analysis links, and build 7 quality inspection models, 48 quality inspection rules, and 76 quality inspection plans. Guarantee the integrity, authenticity, accuracy, timeliness, and standardization of data in the big data platform, and provide effective support for data analysis, decision support, and data mining.
3. Based on the metadata information and the functional panorama of the group, the data classification catalog was constructed. In the first stage, data assets in the domain of human resources, marketing, and supply chain were sorted out, and the portal of the data governance platform realized the automation of data classification and the visualization of data assets. A data classification maintenance plan is established to ensure the accuracy and real-time performance of the classification.
Case 2: Data quality management project of a power group
In order to quickly solve the problems existing in the existing system, such as abnormal flying codes, backtracking of table codes, missed collection by station areas/users, and daily overcapacity, a data quality management platform is planned in the data service platform to quickly locate problem data and perform data analysis on the data. Regular inspections are carried out to improve data quality and provide an effective data basis for subsequent data analysis.
The data service platform builds data source management platform, data integration platform, data authority management platform, data quality management platform, data buffer, data analysis platform, statistical external service interface and other modules in phases. Data quality, providing relevant data support services for different users.
Among them, the data quality management part uses Yixin Huachen data quality management tools to comprehensively improve data quality and reduce user decision-making deviation and loss caused by unreliable data. At the same time, it provides users with a comprehensive quality inspection plan in accordance with the process of specifying the inspection data source, defining rules, regular inspections, and pushing quality inspection reports.
Case 3: Master Data Management Project of Nanshan Group
It is determined that Nanshan Group will establish a set of material master data management model for the whole life cycle. This plan is based on data standards and systems as the cornerstone, and uses management organization, process and platform as the means to achieve comprehensive and efficient management of material master data.
Complete the construction of six major categories of master data, including human resources, finance, procurement, marketing, indicators and other foundations, including material master data, involving 40+ major categories, 3000+ subcategories, 100,000+ entity data, and realize supply chain and other Data docking and sharing of operating systems.
Case 4: Data asset project of Mountain Energy Mining Group
Through the establishment of a group-level big data asset platform, Shandong Energy Linkuang Group uses big data technology to realize data collection, cleaning, analysis and modeling, and realizes the group's full multi-source heterogeneous data collection, and core human resources, finance, equipment, coal Quality and production safety data are used for data governance to form high-quality data assets.
Publish to the whole group through the data asset catalog, and use business metadata to explain the meaning of the data, so that business personnel can find the data they need. Business personnel can apply for the data they need. After the data administrator approves it, the business personnel can use the zero-code agile analysis tool to analyze and retrieve data by themselves, realize data empowerment, and support daily production and operation management.