turboeda 0.2.2

One-command EDA report generator (HTML with Plotly).

turboeda

turboeda is a one-command Exploratory Data Analysis (EDA) report generator.
Give it a CSV or XLSX file — it automatically analyzes the data and creates a polished, interactive HTML report with Plotly charts.

---

📦 Installation

From PyPI

pip install turboeda

From local source (development mode)

# in the project root folder (where pyproject.toml is)
python -m venv .venv
# Windows
.venv\Scripts\activate
# macOS/Linux
# source .venv/bin/activate

pip install -e .

CSV files are read with pandas' default engine. Excel is handled by openpyxl.
CSV encoding is auto-detected via chardet (installed as a dependency).

---

🚀 Usage (CLI)

Short form (recommended):

turboeda "data.csv" -o "report.html" --open
# Excel (auto-uses the FIRST sheet if --sheet is omitted):
turboeda "data.xlsx" --sheet "Sheet1" -o "report.html" --open

Long form (also supported, legacy-friendly):

turboeda report "data.csv" -o "report.html" --open
turboeda report "data.xlsx" --sheet "Sheet1" -o "report.html" --open

Default naming rule:
If -o/--out is not provided, the HTML is saved as <input_basename>_report.html in the same folder as the input file.
Examples:

• C:\data\Iris.csv → C:\data\Iris_report.html
• /Users/me/data/sales.xlsx → /Users/me/data/sales_report.html

Common options:

• --sheet "Sheet1": select Excel sheet if using .xlsx (if omitted, first sheet is used)
• --sep ";" : custom CSV delimiter
• --sample-rows 100000 : sample large files for faster analysis (default: 200000)
• --max-corr-cols 40 : cap number of columns in correlation matrices
• --max-numeric-plots 12 / --max-categorical-plots 12 : limit per-variable charts
• --theme dark|light : choose dark or light theme (default: dark)
• --open : open the generated HTML in your default browser

---

📓 Usage in Jupyter Notebook / Python scripts

from turboeda import EDAReport
from pathlib import Path

# Create and run analysis
report = EDAReport(
    input_path="data.csv",
    theme="dark",           # or "light"
    sample_rows=None,       # use all rows; or set an int to sample large datasets
    max_corr_cols=40,
    max_numeric_plots=12,
    max_categorical_plots=12,
    # Optional UX:
    auto_save_and_open=False,  # if True, saves & opens after run()
    out_path=None,             # custom output name; otherwise uses <input>_report.html
    open_target="tab",         # "tab" or "window" for auto-open
)

results = report.run()

# Export to HTML (default name rule: <input_basename>_report.html)
inp = Path("data.csv")
out = inp.with_name(f"{inp.stem}_report.html")
report.to_html(out.as_posix(), open_in_browser=True)  # set open_in_browser=False if you don't want auto-open

Inline preview in notebook:

from IPython.display import IFrame
IFrame("data_report.html", width="100%", height=800)

---

🎨 Theme

The report supports dark and light themes.
Charts adopt the chosen theme as well (Plotly plotly_dark vs plotly).

• Dark (default):

  turboeda "data.csv" --theme dark
  

• Light:

  turboeda "data.csv" --theme light
  

---

🧠 Notes & Tips

• CSV encoding is auto-detected with chardet; Excel uses the selected sheet (or first if omitted).
• For very large files, consider --sample-rows to speed up initial EDA.
• Datetime detection is heuristic-based and avoids deprecated parsing flags; specify formats upstream if needed.
• On Windows PowerShell, if script activation is blocked, run:

  Set-ExecutionPolicy -Scope CurrentUser -ExecutionPolicy RemoteSigned
  

---

⚡ Requirements

• Python 3.9+
• Packages: pandas, numpy, plotly, jinja2, typer, chardet, openpyxl
  (installed automatically when you pip install -e .)

---

📄 License

MIT License — see LICENSE.