ts-prac

An installable Python package containing three separate Time Series practical workflows:

• Shampoo sales (12-month forecast)
• Daily minimum temperatures (30-day forecast)
• Daily female births (30-day forecast)

Each workflow includes data loading and cleaning, EDA, summary statistics, visualizations, ADF stationarity testing, first-difference demonstration when needed, chronological 80/20 split, ARIMA(1,1,1) evaluation, MAE/RMSE, pickle model persistence, future forecasting, confidence intervals, and interpretation.

Install locally

From this project folder:

py -m pip install .

For development/editable installation:

py -m pip install -e .

Required CSV files

Put the matching dataset CSV files in the current working directory before running:

• shampoo_sales.csv
• daily_minimum_temperatures.csv
• daily_total_female_births.csv

The package writes charts, metrics, and .pkl files to an outputs/ folder in the current working directory.

Run a practical

ts-prac shampoo
ts-prac temperature
ts-prac births

Run all three sequentially:

ts-prac

Or from Python/Jupyter:

from ts_prac import shampoo, temperature, female_births
shampoo.main()
# temperature.main()
# female_births.main()

Build and validate before publishing

py -m pip install --upgrade build twine
py -m build
py -m twine check dist/*

Test on TestPyPI first, then publish to PyPI with Twine. Package name ts-prac must be available on PyPI before you can publish it. Distribution name uses a hyphen, while the Python import name is ts_prac with an underscore.