Skip to main content

Command Palette

Search for a command to run...

Laziest Roadmap for Data Science

This article is incomplete, will be updated soon.

Published
1 min read
B

I am a nerd who has used JS, TS, React.js, Redux ToolKit, Express.js, MongoDB, Node.js, PostgreSQL, Next.js, VITE, Postman, Docker, Linux, Bash, Python, and Langchain.

This is why the line "JACK OF ALL, MASTER OF NONE" fits perfectly. I am sharing my experience with these techs in blogs. starting with JS...

DS Parts:

  • Data Cleaning & Preprocessing || Pandas

  • Data Visualization || Matplotlib

  • Data Analysis

  • Modeling (Machine Learning) || ScikitLearn/SciPy

  • Deep Learning || PyTorch

Python

  • Lists

  • Sets

  • Tuple

  • Functions

Python’s built-in Data Structures and methods are not optimized for working with huge data, and also performance issues there, so we need some data structures that are performant and optimized for huge Data Operations. And Numpy simply does that.

Python One Shot, Learn only the listed Topic, other things, you can learn on the go =>

Numpy

  • 1D, 2D Arrays and Methods

  • Matrices

  • Tensors

Numpy Quick Video Guide =>

Pandas || Data Loading, Cleaning, Preprocessing

Matplotlib || Data Visualization Toolkit

ScikitLearn/Scipy || ML

  • Linear regression

  • SVM

  • Trees, Random Forest

  • k-Means, PCA, etc.

  • Pipelines, preprocessing

PyTorch/TensorFlow || Deep Learning

  • Neural networks

  • Backprop

  • GPU acceleration

  • CNNs, RNNs, Transformers

Gymnasium (OpenAI Gym) => RL Env

Stable Baselines3 => RL Algorithms