This is the lookup site. When a notebook elsewhere in the collection reaches for polars or pyecharts or asyncio, the page explaining it is here.
The notes are deliberately uneven in depth. A few pages are exhaustive because the library repays it, with NumPy, Pandas, and the Python class model each running to hundreds of cells. Others are a page of the commands actually used. Depth tracks how often the thing is reached for, not how important it sounds.
Numerical Core
| NumPy |
The largest page on the site, working through arrays, indexing, broadcasting, and the operations built on them |
| SciPy |
The scientific layer above NumPy |
| NumExpr |
Faster evaluation of numerical expressions through multithreading, for when arrays get large enough to matter |
Dataframes
| Pandas |
The long reference |
| Polars |
The Rust-backed alternative |
| Dask |
Parallel computing that scales from one machine to a cluster, for datasets larger than memory, integrating with NumPy, pandas, and scikit-learn |
Plotting and Diagrams
| Matplotlib |
The default, and what everything else is measured against |
| ECharts |
The most detailed of these, on pyecharts. ECharts describes a chart as one JSON-like option object covering data, axes, series, styling, and interaction. This is the charting library the deep learning notebooks are required to use |
| Plotly |
Interactive publication-quality charts inside Jupyter |
| Bokeh |
Another interactive browser-rendered option |
| Altair |
Declarative interactive plots |
| Graphviz |
Graph and diagram layout from a text description |
| Mermaid |
Diagrams as markdown, which is what the diagram pages in Software Tools are built from |
| Manim |
Programmatic mathematical animation |
The Language Itself
| Python Classes |
The object model at length: definition, inheritance, the dunder methods, and the protocols they implement |
| asyncio |
A complete reference targeting Python 3.11 and up, which is the meaningful floor because TaskGroup, asyncio.timeout(), and task.uncancel() all landed there and they change what correct code looks like |
| Multiprocessing |
The other concurrency route, for when the work is CPU-bound |
| Logging |
The standard library logging module |
| argparse |
Command line arguments |
| Dataclasses |
The decorator that writes __init__, __repr__, and __eq__ for you |
| Datetime |
Dates, times, timezones, and formatting |
| AST |
Parsing, inspecting, and transforming Python source as a tree |
Environments and Packaging
Six tools that solve overlapping versions of the same problem, in roughly the order they appeared.
| venv |
The one in the standard library |
| Pipenv |
Dependency resolution with a lockfile |
| Poetry |
Dependency management plus packaging |
| Conda |
The scientific-stack option, which handles non-Python dependencies |
| uv |
The Rust-based one this collection actually uses |
| direnv |
Loading and unloading an environment automatically on directory change |
| Environment Variables |
How they work in Linux and how to manage them |
Code Quality and Testing
| Ruff |
The fast Python-native linter that replaces flake8 and pylint, and overlaps Black |
| Ruff Rule Categories |
What each rule family prefix means, F, E, W, N, B, and the rest, and why you would enable it |
| Black |
The uncompromising formatter that ends style arguments by making the decisions for you |
| pytest |
Basics through fixtures, plugins, and practice |
| Monkey Patching |
What it is, when it is justified, the risks, and how it shows up in pytest |
Notebooks and Terminal Output
| nbdev |
The framework this entire collection is built on, turning notebooks into libraries and docs |
| Marimo |
The reactive notebook alternative, and how it differs from JupyterLab |
| ipywidgets |
Interactive controls inside a notebook |
| Rich |
Terminal output worth looking at: tables, progress, syntax highlighting |
| tqdm |
A progress bar around any loop |
Data In
| Requests and HTTPX |
The standard HTTP client against the one that adds async and HTTP/2 |
| BeautifulSoup4 |
Parsing HTML and XML, mostly for scraping |
| duckduckgo-search |
Programmatic search across words, documents, images, news, and maps |
| Kaggle |
Pulling competition and dataset files |
| SQLAlchemy |
The 2.x-style guide to Python’s ORM and query builder |
| pytube |
Downloading video |
Odds and Ends
| Latexify |
Rendering a Python function as the LaTeX equation it implements |
| Mito |
A spreadsheet interface over a dataframe |
| pivottablesjs |
Interactive pivot tables in a notebook |
Not Covered Yet
- The site title says MLtools but the content is general Python. Very little here is machine learning specific; the modelling libraries live in ML Methods and DL Methods.
- No page on
typing, despite type hints being unavoidable in modern Python.
- Nothing on
pydantic, which is the validation layer most projects reach for.
pathlib and file handling have no page, only incidental use elsewhere.- Six environment tools, no recommendation. The pages describe each without saying which to pick, though the collection itself uses uv.
- uv is five cells against a tool the whole repo depends on, so it is the thinnest page relative to its importance.
Back to top