How to develop the skill of splitting entities into abstractions?

Question

IT education

I'll give an example:

• Let's say we have a kitchen faucet. Personally, I managed to highlight the following characteristics from him:
  
• Throughput - liter/hour
  
• Mixer type - two-valve / single lever
  
• steel grade
  
• Dimensions - peak height
  
• Manufacturer
  
• Claimed service life
  
• Price
  

An example closer to programming:
A program for "cleaning" data in files of various formats - txt, xlsx, accdb
Requirements that I managed to formulate and that the system should have:

• The whole system should be divided into 4 separate classes-blocks:
  

• Class for reading data, regardless of the format:
  

1. Directly, reads data into the container at the given separator (for txt)
   
2. Allows you to conveniently and flexibly work with the file system - recursively bypassing directories, reading and searching only for files of a given type/-s
   

• Class for data preprocessing and normalization:
  

1. Delete empty entries.
   
2. Bring all lines to the same form.
   

• A class for working directly with prepared data, which is fed a list of "jobs" for each element of the container - for example, a list of pointers to methods that will be applied to certain elements.
  

• A class for convenient creation of a table by passing one of the class methods to the sql query method as a string.
  

These are the examples I'm talking about.
How to learn to immediately analyze the task or problem in a similar, engineering way?

Answer 1

A familiar state, encountered him at the dawn of a pogromist.
Answer: practice. More practice. Even more practice.
Any training and brain training is a mechanical repetition of certain cases, there is no new technique.

Answer 2

Program for "cleaning" data in files of different formats - txt, xlsx, accdb

1. Give the definition of the final result - "cleaned" data.
2. What needs to be done in advance in order to be able to implement the current step on the previous one?
Those. analysis is done in the opposite direction: from the specifics (from the ultimate goal of the task) to the set of abstractions (to the formulation / formulation of the original task).

Answer 3

>Personally, I managed to identify the following characteristics from him:
a meaningless set of parameters ;
we work with models
; any model is one of the possible representations of a real object,
i.e. the model by definition is limited
and the selected set of model properties depends only on the task
, so the manufacturer and throughput are not needed to account for scrap metal. The
second part of the reasoning is the application architecture, MVC, MVP and all that, keywords - layers and area of ​​responsibility