Data analysis, thinking, law breaking and professional knowledge -- analysis method (I)

Analysis method

5W2H Analysis method

What is it? ？

• 5W

  • what( What is it? )
  • when（ when ）
  • where( And where )
  • why( Why? )
  • who( Who is it? )

• 2H

  • how( How do you do it? )
  • how much( How much )

What's the usage? ？

• Can solve simple problems

How to use ？

• Instance of a

  • How to design a product

    • what( What is it? )： What kind of product is this
    • when( when )： When do I need to go online
    • where( And where )： Where to launch these products
    • why( Why? )： Why do users need it
    • who( Who is it? )： Who is this designed for
    • how( How do you do it? )： How this product needs to work
    • how much( How much )： Does this product have the function of paying ？ What's the price ？

• Example 2

  • Design a APP Questionnaire of , How to design the questions on the questionnaire

    • what( What is it? )： You use this APP What to do
    • when( when )： When do you usually use this APP
    • where（ And where ）： What scenarios would you use this APP
    • why( Why? )： Why do you use this APP
    • who( Who is it? )： If you think you like this product , Who would you recommend
    • how( How do you do it? )： What new features do you think we need to add
    • how much( How much )： If you think this APP It helps you , How much money will you pay to buy APP The service in

matters needing attention

• It doesn't work in the face of complex business problems , There are many reasons

Logic tree analysis

What is it? ？

• Disassemble the complex problem into several simple sub problems , Then spread out like a branch

What's the usage? ？

• Simplify complex problems

How to use ？

• Number of tuners （ Fermi problem ）

  • The total working time of all piano tuners in a year

    • How many pianos are there

      • How many people are there in all
      • Proportion of people with piano

    • How many times does the piano tune every year

      • annually

    • How long does it take to adjust once

      • 2 Hours

  • /

  • A tuner works for... Hours a year

    • A year 50 A few weeks
    • Work week 5 God
    • Work every day 8 Hours

matters needing attention

• Logical analysis method is used to solve business problems , Often not alone , Will integrate other analysis methods , Assist in problem solving

Industry analysis methods

What is it? ？

• PEST Analysis method

  • policy （policy）

    • national policy
    • law

  • economic (economy)

    • National income
    • The income level of consumers

  • social (society)

    • Regional population 、 Age
    • Buying habits
    • educational level

  • technology (technology)

    • The impact of external technology on the development of the company

What's the usage? ？

How to use ？

matters needing attention

Multi dimensional analysis method

What is it? ？

• dimension = angle
• Take apart = dimension 1+ dimension 2+ dimension 3

What's the usage? ？

• Simpson paradox

  • Looking at the data as a whole and looking at different parts of the data will lead to the opposite conclusion

  • example —— Choose a hospital to see a doctor

    • A Hospital admission 1000 Patients ,900 Cases survived

      • A The hospital 100 Severe cases ,20 Cases survived
      • A The hospital 900 Mild cases ,20 People have been killed

    • B Hospital admission 1000 Patients ,800 Cases survived

      • B The hospital 400 Severe cases ,200 Cases survived
      • B The hospital 600 Mild cases , All survive

How to use ？

• Disassemble from the index composition

  • Store sales

    • New user sales

      • Sales of new users = Number of new users Conversion rate New customer unit price

    • Sales of regular users

      • Sales of regular users = Number of old users Repeat purchase rate Regular customer unit price

• Disassemble from business process

  • Promotion effect （ Disassemble according to the index composition ）

    • Subdivide by Region

      • Users in first tier cities are more interested in the company's products

    • Breakdown by gender

      • Male users are more interested

    • Subdivide by channel

      • channel A The largest number of users

      • channel B Close to A

      • channel C least

      • Which channel has higher quality

        • Disassemble according to the business process

          • See channel advertisements
          • Be attracted to enter the store by advertisements
          • Choose products of interest in the store
          • Choose good products , Finally decided to buy

matters needing attention

• When examining the data, you may not notice the differences in the internal composition of the data by looking at the data as a whole , Lead to Simpson paradox

After disassembling according to the business process, it is found that although the channel A Bring more users , But the number of buyers is lower than the channel B Number of users brought , So channels B Better user quality