How to start Data Science

Start with learning Python. If you have previous experience on coding, that's good. Knowledge transfer. If not, just learn it. Learn from the basic until function definition is enough. Want to go extra? Learn until OOP.

Write the code on notebook. Start with Jupyter Notebook. You can setup with Anaconda (get full package of everything), or small setup with Miniconda. Easy to use. Learn how to run python script after that.

Learn to use Pandas. Learn Pandas so that you can do all data manipulation in the notebook. So you don't need to open your Excel to do something. When I said manipulation, there are tons of manipulation stuff you can do.

Learn about modelling. Start with Scikit-learn. The package also has data manipulation feature too. From sklearn, learn about encoding, modelling (regression, classification), cross validation, kfold and so on. Here is the starting point of learning about machine learning.

Add your tool by learn deep learning too. Use Pytorch framework.

Learn Matplotlib and Seaborn. You can see and analyze better with visual.

Learn about hyperparameter tuning library such Optuna. Once you got the base model, automate the process of getting the best model with Optuna. Make coffee, sit down, train the model and let Optuna find the best model. Save and track the model with MLFlow. Best combo ever.

Learn about Shapley. It is a model interpretability method. If your boss or lecturer asking you why the model decided to choose the class, you can explain it smoothly.

Change your Googling style. Use Google dorking. Start seeking information from Kaggle. Here is the place where all grandmaster and people who crazy on machine learning gathered. Use inurl:kaggle how to ... while Googling.

Utilize Kaggle. Join Kaggle competition. Click the Competition, go to Community to start. Kaggle also provide free courses by chapter that we can enroll (located on Learn tab).

Keep learning.