Data Science & Machine Learning Roadmap

Statistics-first path to building and honestly evaluating your first models.

10 milestones · ~177 hours total

1. 1

   Python, NumPy and pandas

   ~25h

   Core Python, arrays, DataFrames, indexing, grouping and merging.

   Practice this →
2. 2

   Statistics and probability

   ~20h

   Distributions, mean/variance, sampling, hypothesis testing and confidence intervals.

   Practice this →
3. 3

   Data cleaning and exploratory analysis

   ~15h

   Missing values, outliers, types, and asking the right questions of a dataset.

   Practice this →
4. 4

   Data visualisation

   ~8h

   Charts that communicate: matplotlib/seaborn, choosing the right plot.

   Practice this →
5. 5

   Supervised learning

   ~25h

   Linear/logistic regression, trees, ensembles and k-NN with scikit-learn.

   Practice this →
6. 6

   Model evaluation

   ~12h

   Train/validation/test splits, cross-validation, precision/recall/F1, ROC and data leakage.

   Practice this →
7. 7

   Feature engineering and pipelines

   ~12h

   Scaling, encoding, selection and reproducible sklearn pipelines.

   Practice this →
8. 8

   Unsupervised learning

   ~10h

   Clustering, PCA and anomaly detection.

   Practice this →
9. 9

   Intro to deep learning

   ~25h

   Neural networks, backpropagation, and a first model in PyTorch or TensorFlow.

   Practice this →
10. 10

    Projects and deployment

    ~25h

    Two end-to-end projects with a clear question, honest evaluation, and a simple deployed demo.

    Practice this →