Machine Learning & Engineering
Tested notes on machine learning, Python, systems, and practical development.
2026
- AI code review vs unit tests: what catches which bugs?
- Polars 2.0 upgrade: the lazy join that now returns rows in a different order
- Python 3.15 frozendict: when to use it, and where pandas still expects a dict
- Target encoding leakage: what scikit-learn’s TargetEncoder cross fitting protects you from
- Your first LSTM forecast in PyTorch: from CSV to predictions
- How to choose a classification threshold from error costs (scikit-learn TunedThresholdClassifierCV)
- Recursive vs direct multi-step forecasting: what a one-step score hides
- KFold vs TimeSeriesSplit: how shuffled cross-validation hides time-series leakage
- Bit tricks in Python, one bit at a time: why ~5 is -6 and how to get 8-bit results
- When does a greedy algorithm fail? Coins, meetings and knapsacks, checked against the optimum
- Dynamic programming, step by step: Fibonacci, LCS and 0/1 knapsack in one app
- BFS, DFS, Dijkstra and A* on the same grid: what each one actually guarantees
- Before replacing LightGBM with TabPFN-3.5, check your evaluation split
- Before benchmarking free-threaded Python, check what NumPy already does
- Loop vs vectorization: how much faster is NumPy, really?
- Python 3.15 for data work: lazy imports, frozendict and a new profiler
- How to use MinMaxScaler in scikit-learn: fit, transform, and inverse_transform
- LightGBM LambdaRank in Python: group, NDCG, and the mistakes that don’t raise an error
- Can you trust an AI agent’s logs? The viewer is part of the attack surface
- Sorting algorithms, visualized in Python: insertion, bubble, selection and quick sort
- Self-attention from scratch in NumPy (checked against PyTorch)
- LSTM input shape explained: (batch, timesteps, features)
- pandas vs Polars vs DuckDB in 2026: a real benchmark on 10 million rows
- Why fit preprocessing on training data only? A data leakage demo
- How to show a progress bar in GridSearchCV (scikit-learn 1.9 callbacks)
- MinMaxScaler vs StandardScaler: which scaler should you use?
2023
- KoAlpaca 에 대해 공부 - 3
- Self Attention에 대해 공부
- RIPPER Algorithm 이란?
- Mojo가 모죠? - 2
- KoAlpaca 에 대해 공부 - 2
- KoAlpaca 에 대해 공부 - 1
- Python으로 PDF를 페이지별 JSON으로 변환하기 (pdfminer.six + KoNLPy)
- Python의 비밀! Python의 속도를 빠르게 만드는 방법
- Python 문법을 고급스럽게 바꿔보자.
- Mojo 언어란? 설치부터 파이썬 속도 비교까지 (2026 업데이트)
- Tensorflow의 Dataset 사용법 정리
- cuDNN을 설치해보자
- LSTM Input Shape 정리 (input_shape, batch_input_shape, Incompatible shapes 에러)
- torch.cuda.is_available() False 해결 (no kernel image 에러 포함)
- RNN에 관한 정리?
- TensorFlow란?
