To become a data scientist, a solid foundation in machine learning (ML) is essential, as it is a core component of the role. You need to understand key concepts such as supervised and unsupervised learning, model evaluation, feature engineering, and common algorithms like regression, decision trees, and neural networks. Proficiency in implementing ML models using libraries like Scikit-learn, TensorFlow, or PyTorch is also crucial. However, data science is broader than just ML; it includes data wrangling, statistical analysis, data visualization, and domain expertise. While you don’t need to be an ML expert to start, a strong grasp of ML concepts and practical experience in applying them to real-world problems is necessary to excel as a data scientist.

I would be happy if you read my articles and recommend them to increase their visibility.

Both are serving each other. you have to master both.

1. Essential ML Skills

• Supervised & Unsupervised Learning: Regression, decision trees, clustering, dimensionality reduction.
• Model Evaluation: Precision, recall, F1-score, cross-validation.
• Feature Engineering: Handling missing data, scaling, encoding.

2. Intermediate Skills

• Deep Learning Basics: CNNs, RNNs, transformers.
• Optimization & Hyperparameter Tuning: Gradient descent, grid search.
• Deployment: Flask, FastAPI, Docker.

3. Advanced (For Research Roles)

• Reinforcement Learning, GANs, AutoML
• Bayesian ML, Probabilistic Graphical Models
• ML Interpretability (SHAP, LIME)

f you define "how much" in terms of theoretical knowledge, my guess is that for someone that writes machine learning algorithms the answer is "an in depth knowledge of the underlying mathematics and algorithms". For someone that uses them the answer is "a firm understanding of building and validating models". All should know the basic tasks in supervised, unsupervised and dimension reduction though.

The "type" of scientist/analyst is quite relevant too. I expect a statistical learning approach by an analyst. It is quite similar to machine learning but falls more in the interpretable side. A data scientist that falls deep in the machine learning side is likely to be much more interested in deep learning and algorithm performance.