当前位置:网站首页>The boundary of Bi: what is bi not suitable for? Master data, Martech? How to expand?
The boundary of Bi: what is bi not suitable for? Master data, Martech? How to expand?
2022-07-07 20:05:00 【Bi visualization of Parker data】
When we communicate with customers at some customer sites , Customers often mention some scenarios and problems , Ask if you can BI It's inside .
BI Not good at what
For example, yes BI after , All data is extracted from the data source to BI Layer processing , Data processing is very standard , But the data in the business system is still not standardized , The problem of data quality has not been really solved . There is also a BI after , I want to use some data to push some information to qualified users 、 Precision marketing , Users' feedback after receiving the information also hopes to precipitate to BI In the data warehouse , And so on .
BI - Parker data BI Visual analysis platform
In fact, strictly speaking , None of this is BI Areas of expertise .BI What I'm good at is still getting through all kinds of data systems , And pass BI Data warehouse modeling builds all kinds of reasonable analysis models , Externally, it provides decision support to the management in the form of visual reports , Or through in-depth BI Analyze and position problems in enterprise business and management , Whether to make a decision in the end 、 What kind of decision to make is still up to people .
So the two scenes just mentioned are actually located in BI Upstream inlet end of , One is located in BI Downstream outlet end of .
BI Upstream data problems
Data quality problems of data source BI It can't be solved ,BI What can be solved is to pull the data from the data source into BI In the data warehouse , adopt ETL Complete data consolidation 、 cleaning 、 Governance work , Will be nonstandard 、 Non analyzable data becomes normative 、 Analyzable data forms support front-end visual analysis .
BI Data warehouse - Parker data BI Visual analysis platform
For example, we often see in some enterprises , The planning of information system is carried out in stages , Different systems and departments are using , Due to the lack of certain unified specifications in advance , In addition, different business departments use different systems, such as product coding 、 The maintenance of supplier file information is inconsistent , The result is BI When modeling data retrieval, you don't know which system's dimension file data is the main data , It takes a lot of time 、 Focus on the consolidation of such data 、 Development work such as elimination of dirty data .
therefore , For the quality control of these data 、 The cleaning of dirty data is set to BI Level to deal with . If preprocessing is needed , You need to rely on the master data system 、 The data governance system performs processing control on the data source , Establish unified master data file information , All dimension files are uniformly controlled by the master data system , Then distribute it to all business systems to establish unified data standards .
Or establish data archive standards when planning and building each business information system , Follow certain data rules , This is also data quality 、 Data archive information 、 The front of data specification . Therefore, this kind of data processing is either advanced to BI Solve it before , Or post to BI To deal with . But in BI To deal with ,BI It can only control the quality of data exported , It is impossible to trace back to the data source to solve the data quality of the data source 、 Data standards .
BI Downstream output problems
What I just mentioned is BI Upstream data problem , Let 's talk about it again BI The downstream output of , For example, at the beginning, we should do some user clustering 、 Do some precision marketing , For example, in the field of marketing MarTech. In fact, its processing flow is also to collect various internal systems 、 The data of the external environment is processed , Build some models according to certain logical calculation rules , Then use these data to do some marketing work according to certain scenarios , This part is not BI What you are good at .
Visualization screen - Parker data BI Visual analysis platform
BI As we have just mentioned, it is when the internal data warehouse completes the analysis model , Responsible for front-end visual analysis . That is to say, the front-end visualization is over , It will not be involved in the subsequent business process processing , For example, marketing promotion is a business process . But actually look at the data processing process just now , and BI Of ETL The process is also very similar , All kinds of databases need to be collected 、 Interface, whether external or internal data , It also requires very complex data processing , Build a model .
BI Expand capabilities
According to BI Analysis of the problem , We will find that BI If you want to take a step further , In fact, it is the link of data service , Involve in subsequent business links by providing external data services , That's it BI Connect and connect with subsequent business processes . The application field of this scenario is very broad , Like the marketing just mentioned 、 Intelligent replenishment in the retail industry 、 Intelligent early warning of inventory, etc .
BI Expand - Parker data BI Visual analysis platform
therefore ,BI The upstream and downstream expansion capacity of the field is still very strong , Its upstream and downstream ends are business systems , One is a business system that provides data sources , One is to enter the downstream business system through normative data , And in the middle is BI The core processing link .
Upward ,BI It can be extended to master data 、 Data governance . Down ,BI Data application scenarios that are expanded to various sub fields through data services , But its industrial characteristics are relatively strong . These are the future BI A direction that can be developed .
actually , Let's stop here , Do you also feel that it is somewhat similar to the current data center , Three links : Data collection 、 Data processing 、 Data services . therefore , Many technical concepts are basically the same , Just their positioning 、 Bias 、 Industry 、 Different business forms different solutions to solve some specific scenario problems .
边栏推荐
- 831. KMP string
- Sword finger offer II 013 Sum of two-dimensional submatrix
- R语言ggplot2可视化:使用ggpubr包的ggecdf函数可视化分组经验累积密度分布函数曲线、linetype参数指定不同分组曲线的线型
- Notes...
- Cloud 组件发展升级
- mock. JS returns an array from the optional data in the object array
- R language uses ggplot2 function to visualize the histogram distribution of counting target variables that need to build Poisson regression model, and analyzes the feasibility of building Poisson regr
- 项目经理『面试八问』,看了等于会了
- 841. 字符串哈希
- Introduction to bit operation
猜你喜欢
Leetcode force buckle (Sword finger offer 36-39) 36 Binary search tree and bidirectional linked list 37 Serialize binary tree 38 Arrangement of strings 39 Numbers that appear more than half of the tim
CSDN语法说明
谷歌seo外链Backlinks研究工具推荐
openEuler 资源利用率提升之道 01:概论
关于cv2.dnn.readNetFromONNX(path)就报ERROR during processing node with 3 inputs and 1 outputs的解决过程【独家发布】
el-upload上传组件的动态添加;el-upload动态上传文件;el-upload区分文件是哪个组件上传的。
mock.js从对象数组中任选数据返回一个数组
Sword finger offer II 013 Sum of two-dimensional submatrix
转置卷积理论解释(输入输出大小分析)
开源OA开发平台:合同管理使用手册
随机推荐
How to buy stocks on your mobile phone and open an account? Is it safe to open an account
[RT thread env tool installation]
JVM class loading mechanism
ASP.NET幼儿园连锁管理系统源码
时间工具类
torch.nn.functional.pad(input, pad, mode=‘constant‘, value=None)记录
The project manager's "eight interview questions" is equal to a meeting
LeetCode力扣(剑指offer 36-39)36. 二叉搜索树与双向链表37. 序列化二叉树38. 字符串的排列39. 数组中出现次数超过一半的数字
Ucloud is a basic cloud computing service provider
微信公众号OAuth2.0授权登录并显示用户信息
pom.xml 配置文件标签作用简述
Ways to improve the utilization of openeuler resources 01: Introduction
我的创作纪念日
Time tools
编译器优化那些事儿(4):归纳变量
国家网信办公布《数据出境安全评估办法》:累计向境外提供10万人信息需申报
R语言fpc包的dbscan函数对数据进行密度聚类分析、查看所有样本的聚类标签、table函数计算聚类簇标签与实际标签构成的二维列联表
ASP. Net kindergarten chain management system source code
毕业季|遗憾而又幸运的毕业季
九章云极DataCanvas公司摘获「第五届数字金融创新大赛」最高荣誉!