Multidimensional pivoting analysis of CDA level1 knowledge points summary

Strategic analysis

This chapter is a new addition to the new edition , It accounts for a large proportion in the exam , It's difficult to learn , We need to pay high attention to .

Intensive lecture at the examination site

Overview of multidimensional data model OLAP

For table data structures , Take fields or records as references to data 、 Data of basic units of operation and calculation .
dimension – Business perspective , Dimension fields – The text type , Measure — Business behavior results , Measurement field – Numerical type .
Pivoting – Summarize and analyze scattered data .
Dimension table – Only contain dimension information , Fact table – It contains dimension information , It also contains measurement information .
Multidimensional data model is also called cube 、 Cube , It refers to data sets of different categories that are related to each other through some kind of connection . The advantage is that you can comprehensively map the actual situation of a business with data from multiple perspectives .

Multidimensional data model creation method

Connection summary of two adjacent tables , I. list dimensions , A table shows the measurement , Select summary calculation rule .

For one-way filters , The starting side of the arrow is the dimension 、 Pointing to one side is a measure .OLAP When connecting summary , Who measures , Who is the main watch （ Who knows the business result data , Who is the boss ）

Three corresponding relationships ：
1. one-on-one . It hardly appears in the actual scene . The primary key is connected with the primary key 、 Two tables have the same primary key .
2. Many to many . There will be , But try to avoid . Non primary key connection non primary key , It will double the measurement .
3.1 For one more （ A one-way ）. Try to use many to one . When one-way filtering direction , One table filters multiple tables .
Type1: I. list dimensions , Multi table measurement .
Type2: error

3.2 For one more （ two-way ）. When two-way filtering direction , Multiple tables can filter one table , However, the filtering method is different from that of filtering multiple tables with one table .
Type1: I. list dimensions , Multi table measurement .
Type2: A table shows the measurement , Multi table dimensions .

Cross table filtering
Cross connect Multiple connection modes exist at the same time , But only one model works .

Three models （ Compulsory examination ）：
Star pattern – A fact table is connected with multiple dimension tables
Snowflake mode – Based on the star pattern , Dimension tables are linked to more dimension tables
Constellation mode – Several fact tables share some dimension tables

5W2H Thinking model

Basic perspective rules

5 There are three basic rules of perspective ：
total sum、 Count count、 Average average、 Maximum max、 minimum value min
Average trap ： use average The total average value obtained is the total average value of the primary key .

Perspective rule extension

5 Comparison calculation rules ： Average ratio （ Actual value and average value ）、 Benchmark ratio （ Actual value and reference value ）、 The goal is better than （ Actual value and target value ）、 Standard ratio （ Actual value and standard value ）、 Proportion （ Part and total 、 Proportion of sales in different regions ）.
5 Summary rules in three times ：MTD Summary from the beginning of the month to the current date 、QTD Summary from the beginning of the quarter to the current date 、YTD、 Chain ratio 、 Year on year . Among them, month on month and year-on-year are suitable for long-term data indicators .
2 Comparison summary rules ： Comparison percentage （ actual value / Contrast value *100%）、 Percentage difference –（ actual value - Contrast value ）/ Contrast value *100%.
Inter line perspective ： Calculate the overall summary value for each row of values , Similar to windowing function .

Multidimensional perspective analysis application

Exercise summary