NumPy & Data Wrangling

Scaling, encoding, splitting and reshaping data the way training pipelines actually need it.

Beginner · 18 problems · 0 passed · Background reading: Python Engineering for AI — Overview

1. 01NumPy Vectorized Min-Max Feature Scalingeasy
2. 02NumPy Vectorized Z-Score Standardizationeasy
3. 03NumPy Vectorized One-Hot Matrix Encodingeasy
4. 04Vectorized Softmax Normalizationeasy
5. 05NumPy Vectorized Confusion Matrixeasy
6. 06Pandas GroupBy Metric Aggregationeasy
7. 07Pandas Missing Value Imputation Pipelineeasy
8. 08Vectorized Numerically Stable Log-Sum-Exphard
9. 09Vectorized Euclidean Distance Matrix via Gram Matrixhard
10. 10Stratified Train-Test Split Partitioneasy
11. 11Streaming Moving Variance (Welford)medium
12. 12Vectorized Batch Collator with Sequence Paddingmedium
13. 13Vectorized Multi-Head Attention Reshape and Transposemedium
14. 14Time-Window Event Stream Sessionizationmedium
15. 15User Cohort Retention Matrix Enginemedium
16. 16Vectorized Layer Normalization with Learnable Affinehard
17. 17Vectorized Causal Triangular Attention Maskinghard
18. 18FlashAttention Online Softmax Block Accumulatorhard

Next track: Math for ML