PYGuide — Python Reference Cookbook

Runnable Python snippets with their output written inline, covering language basics through pandas, plotting and scikit-learn.

• Runnable snippets — one file per topic, each a flat list of independent examples.
• Verified inline outputs — every result is recorded as a # Output: comment, so you can read the answer without running anything.
• Basics → data analysis → ML — language fundamentals, data structures, NumPy, pandas, Matplotlib, Seaborn, scikit-learn, and SQLite.
• A personal reference, not an application or a course — there's no entry point and no CLI.

Overview

This is a personal reference cookbook of Python snippets — my own notes, kept as short, readable examples with the expected result written inline as a comment. It is not a tutorial and not an application. Each file covers one topic as a flat list of independent snippets, meant to be looked up rather than worked through in order.

What Makes It Different

• Files are designed to run top to bottom where practical. Most snippets are plain expressions whose result is recorded in a comment rather than printed, so running a file often produces little or no console output — but the files execute cleanly end to end rather than depending on missing local data.
• Outputs are written inline as # Output: comments. The result travels with the code, so the snippet is useful while reading a diff, a search result, or a single line pasted elsewhere.
• Version-dependent output is marked, not glossed over. Where a result depends on the installed library version — dtype names, scalar repr, sort order — the comment says so instead of presenting one version's output as universal. See Verified Environment.
• Self-contained data. Examples use either the small fixtures in data/ or seaborn's built-in datasets, so nothing depends on files that only exist on one machine.

How to Use

• Open the file for the topic you want (see the index below) and read top to bottom.
• Every snippet shows its result as a # Output: comment — you can read the answer without running anything.
• To find how a specific method or function is used, search for it by name across files (grep, your editor's project search, etc.) — that's the intended way to navigate this repo, not a table of contents.
• A few files load small fixture files from data/, or seaborn's built-in datasets (tips, titanic) — see Data Files below.

Topic Index

Each row's description is drawn from what the file actually contains, not from its name. Line counts reflect the files as they currently stand.

Language Basics

File	Covers	Lines
Boolean.py	Defining a boolean and checking its type	5
Input.py	Reading user input with input() and building a greeting from it (sample session shown, since the real output depends on what you type)	11
Conditionals.py	Comparison/logical operators, if/elif/else, and membership tests (in/not in) across dict, list, set, and tuple	38
Conversion.py	Converting between str, int, and float	8
Loops.py	break/continue, for loops over lists, dicts, sets, tuples, and DataFrame columns, and while loops	82
Numbers.py	type() on float and int	6
Strings.py	Indexing, slicing, and common string methods (capitalize, count, split, upper)	45

Data Structures

File	Covers	Lines
Dictionary.py	Creating dictionaries, accessing keys/values, .get() with a default, adding entries	26
List.py	Indexing, slicing, comprehensions, and common list methods (append, insert, pop, sort, reverse)	63
Set.py	Creating sets, union, intersection, and converting a list to a set to drop duplicates	25
Tuple.py	Creating tuples, indexing, count(), and index() on an immutable sequence	17

Functions & Structure

File	Covers	Lines
Annotation.py	Variable, function, class, and list type annotations, including the modern list[int] generic and int | str union syntax	49
Functions.py	Default arguments, *args/**kwargs, enumerate, filter, lambda, map, random, range, zip	97
Module.py	Importing a module, a name from it, and a submodule three different ways, backed by a small local stub package (serhatmodule/) so the example actually runs	25
OOP.py	Classes, inheritance, encapsulation, polymorphism, and abstraction via abc.ABC	109
ErrorHandling.py	try/except, else, finally, raise, and custom exceptions	52
Files.py	Writing, reading, and appending to a local text file	20

Data Analysis

File	Covers	Lines
NumPy.py	Array creation, indexing/slicing, arithmetic, matrix operations, random arrays, linear algebra	128
Pandas.py	Series and DataFrame creation, .loc/.iloc, Excel/CSV I/O, missing-data handling, groupby, concat/merge, .apply() — the deepest file in the repo	666
DateTime.py	Converting text dates to pandas datetime64, plus stdlib datetime creation, formatting, parsing, and arithmetic	37

(DateTime.py is grouped here rather than under Language Basics — it's really a pandas + stdlib datetime file, not a core-language topic.)

Visualisation

File	Covers	Lines
Matplotlib.py	Line plots, subplots, figures/axes, histograms, scatter plots, line styling	112
Seaborn.py	Dataset inspection (columns, corr, value_counts, unique) plus bar/box/cat/dis/heatmap/line/scatter/histogram plots on the tips dataset	86

Machine Learning

File	Covers	Lines
BalancingData.py	Upsampling, downsampling, and SMOTE for an imbalanced binary target	89
Encoding.py	One-hot, label, and ordinal encoding of categorical columns	80
FeatureEngineering.py	Handling missing values with mean/median/mode imputation	31
Sklearn.py	A full linear regression pipeline: train/test split, feature scaling, training, evaluation metrics	56

Databases

File	Covers	Lines
SQL.py	SQLite CRUD, SELECT variants (columns/limit/order/where), and aggregate functions (count/avg/max/min/group by) via sqlite3	242

Getting Started

Requires Python 3.10+ (Annotation.py uses list[int] and int | str syntax).

git clone https://github.com/serhatmercan/PYGuide.git
cd PYGuide
pip install -r requirements.txt

• Run scripts from the repository root — files that read fixtures use relative paths such as data/employees.csv, so they will not find their data from another working directory.
• For headless plotting, set the non-interactive Matplotlib backend: MPLBACKEND=Agg python Matplotlib.py. Under Agg the plt.show() calls emit a harmless "non-interactive" warning instead of opening a window.
• Running a few files creates local artifacts — SQL.py → database.db, Files.py → test.txt, Matplotlib.py → age_weight_plot.png. All three are gitignored.

Requirements

Pure standard library, no install needed: Boolean.py, Input.py, Conditionals.py, Conversion.py, Numbers.py, Strings.py, Dictionary.py, List.py, Set.py, Tuple.py, Annotation.py, Functions.py, Module.py, OOP.py, ErrorHandling.py, Files.py, SQL.py.

Third-party:

Library	Used by
numpy	NumPy, Pandas, Matplotlib, Sklearn, BalancingData
pandas	Pandas, DateTime, Loops, Encoding, BalancingData
matplotlib	Matplotlib, Seaborn
seaborn	Seaborn, Loops, Matplotlib, Encoding, FeatureEngineering, Sklearn
scikit-learn	Encoding, BalancingData, Sklearn
imbalanced-learn	BalancingData

Data Files

Every example is self-contained. Data comes from one of two places:

• Small synthetic fixtures committed to data/, generated for these examples and containing no real-world data: city_temperatures.xlsx, city_temperatures_missing.xlsx, employees.csv, employees_1.csv, employees_2.csv (all used by Pandas.py), and support_tickets.csv (DateTime.py).
• seaborn's built-in datasets (tips, titanic), which download nothing and need no local file: Loops.py, Matplotlib.py, Seaborn.py, Encoding.py, FeatureEngineering.py, Sklearn.py.

Module.py's import examples are backed by the small serhatmodule/ package in this repository (__init__.py + serhatsubmodule.py), so the module, name, and submodule imports it demonstrates all resolve for real.

Conventions

• # Output: <value> appears after an expression to show what it evaluates to — read it in place of running the code.
• # Renders: <description> appears after a plotting call (Matplotlib.py, Seaborn.py) instead of # Output:, since a plot has no printable return value — it describes what the call draws.
• # === Section === divides a file into named parts so you can jump to a topic within a file.
• # Behaviour varies by version: ... flags a result that depends on your installed library version (dtype names, scalar repr, sort order) rather than presenting one version's output as universal.

Verified Environment

Inline outputs are checked by running the code, not by inspection. The values recorded in # Output: comments were produced against these library versions:

pandas            3.0.2
numpy             2.4.2
matplotlib        3.10.8
seaborn           0.13.2
scikit-learn      1.8.0
imbalanced-learn  0.14.1

If you're on different versions, expect differences in a few specific places that are called out inline wherever known: dtype names (e.g. pandas 3.x reports dtype='str' for string columns where older pandas reports dtype='object'), scalar repr (NumPy 2.x displays a bare integer result as np.int64(5) in a REPL, not plain 5), and .info() output (memory usage and the printed class path both shift slightly by version). None of this affects the underlying values — only how they're displayed.

Scope & Limitations

• Coverage is uneven by design. Pandas.py is a deep reference; Numbers.py covers only type() on two values. Depth follows what has actually been needed, not a syllabus.
• This is a personal, evolving reference, not a comprehensive Python or data-science course. It will keep changing as more of it gets used and re-checked.
• The machine-learning files are syntax references, not modelling advice. They demonstrate individual steps — encoding, imputation, resampling, a regression pipeline — and each notes inline where a real workflow would order things differently to avoid leaking test data.
• Verification is local script execution only. There is no test suite and no CI in this repository.

License

MIT — see LICENSE.

Author

Serhat Mercan — github.com/serhatmercan