Best Known Methods (BKM) - for Python App development

Afte years of development in python, here are my go-to approaches for building any (desktop/ web) application.

Framework / Structure

+----------------------------+--------------------------------+
|  UI and File Management    | User management (if web-hosted)|
+----------------------------+--------------------------------+
           |              |            
           v              v            
+--------------+     +-------------+   
| Extraction / | --> | Processing  |  
| Compaction   |     |             |   
+--------------+     +-------------+   
           |              |            
           v              v         
+-------------------------------------------+
|       Subroutines / Small Functions       |
+-------------------------------------------+

Extract the Input files when

• Big files requiring heavy libraries (e.g., Excel, SnP, PDF)
• Files needing asynchronous handling (e.g., large CSV files and frequently cross-referenced)
• Files with sparse data (e.g., sparse matrices)
• Extract files into smaller-sized data objects

---

Data Storage

• Extract and store in a lightweight, easily accessible format (e.g., Pickle)
• Trim the data: keep it small, compact, and accurate
• Ensure data writing is reliable and error-free

---

Encapsulation

• Use objects for encapsulating data when:
  • Input and processing are large
• e.g. - classes, import-able files in python

---

Modular Code

• break tasks into as many moderately sized chunks as possible
• Helps in debugging
• Lets user focus on only one section, instead of getting overwhelmed
• use subprocess, functions/subroutines

---

Use Web-Hosted Apps When

• Tasks require small execution time but depend on large libraries
• Libraries are not supported on all PCs (e.g., ARM Windows)

---

Processing Considerations

• If processing takes a long time - Show a progress bar to keep the user informed

---

Use Deterministic Programming When

• Data is sensitive and impacts cost or timeline
• Always ask the user to verify the data

---

When to use AI

• Use AI for:
  • For quick access to tutorials
  • Verify approaches and ideas
  • Understanding and implementing new domains (libraries, languages)
  • For new trends, explore forums and experiment

---