My coding style

Aim of the project

I’m gonna write about how i made my coding experience more structured less mental breakdown. I hope you find it useful.

Structure of the project

first structuring the imports. When the projects getting bigger in every file you have to go through them and import them. This is very time consuming and error prone. So I created a import.py file. This file is just a collection of imports, and it’s used to import the project as a whole.

Folder Structure

After when you start the project i need to create my folder structure. I already created a boilerplate for all of my projects.

It’s a base folder it’s not specified about any specific project, but it’s a good start.

In general DS projects i worked, you have to do some type of the jobs in any case, like data collection, creating storage systems, data cleaning, etc. etc.

So my base folder structure goes like this;

• /dev/projectname

  • .venv
  • .git
  • .github
    • workflows
      • commit.yml
      • main.yml
  • config
    • settings.json
    • requirements.txt
    • pyproject.toml
    • config.yaml
  • data
    • raw
    • processed
    • external
  • docs
    • API.md
    • FAQ.md
    • index.md
    • architecture.md
    • changelog.md
    • CONTRIBUTING.md
    • database_schema.md
    • requirements.md
    • user_guide.md
    • visual_guide.md
  • logs
    • mlrun_logs
  • models
    • lstm_model.h5
    • rnn_model.h5
    • .
    • .
  • reports
  • scripts
  • src
    • init.py
    • pycache
    • analysis
    • data
    • data_collection
    • data_ingestion
    • data_processing
    • database
    • deployment
    • models
    • monitoring
    • optimization
    • preprocessing
    • training
    • utils
    • visualization
    • config.py
    • main.py
  • templates
  • tests
  • README.md
  • LICENSE
  • .gitignore

I’m gonna share my full folder structure with .zip file. You can download the full folder structure here..

Coding Style

In every project you don’t need to know what to write and how to write but you need to understand your idea well. When you understand the logic of your idea you can start to build your code structure. (Not really coding just throwing the idea of your system). In general it looks like this for me:

def data_cleaning():
    insert here import function for your dataset
    check your data types
    drop unnecessary columns
    check for missing values
    fill missing values
    check for duplicates
    save your cleaned dataset
    pass

def data_preparation():
    transform date format
    encode categorical variables
    scale numerical variables

def train_model():
    choose the right model
    split your data into train and test
    train the model
    evaluate the model
    save your trained model

def main():
    data_cleaning()
    data_preparation()
    train_model()
    pass

if __name__ == "__main__":
    main()

Even throwing the ideas like this to a draft document will give you a better understanding how the code works and where do you want to go.