viewdt

Advanced interactive data table widget for pandas DataFrames — Python port of the R ViewR viewdt() function.

viewdt() transforms any pandas DataFrame into a high-performance, fully self-contained HTML explorer with zero browser-side dependencies. Column statistics are profiled in Python before the page renders, so the widget is fast even on large datasets.

---

Features

Feature	Details
Type badges	# numeric   A text   T/F logical   ⏱ datetime
Spark histograms	Inline SVG mini-charts in every column header
Completeness bars	Per-column fill-rate indicator (green / amber / red)
Column labels	Reads series.attrs["label"] — compatible with haven / ADaM clinical datasets
Virtualised grid	Renders only visible rows — handles hundreds of thousands of rows smoothly
Global search	Live cross-column text filter
Visual query builder	Multi-condition AND / OR filtering with type-aware operators
Data Insights drawer	Full histogram, descriptive stats, and category charts — click any column header
Column picker	Toggle column visibility at runtime
Code export	Generates pandas, Python, and SQL that reproduce the current filter state
Dark / light / auto theme	Follows the system preference by default; toggle at runtime
Standalone HTML	One self-contained file, no CDN or internet connection required

---

Installation

pip install viewdt

Python 3.9+ and pandas 1.3+ are required. No JavaScript build step is needed.

---

Quick start

import pandas as pd
from viewdt import viewdt

df = pd.read_csv("sales.csv")
viewdt(df)               # renders inline in Jupyter
viewdt(df).show()        # opens in the default browser
viewdt(df).save("explorer.html")  # exports a standalone file

Or explore one of the built-in sample datasets immediately:

from viewdt import viewdt, load_iris, load_mtcars, load_titanic

viewdt(load_iris())       # 150 rows — flower measurements
viewdt(load_mtcars())     # 32 rows  — 1974 car specs
viewdt(load_titanic())    # 891 rows — survival data with missing values

---

API reference

viewdt(data, options=None, dataset_name=None)

Create an interactive data-explorer widget from a pandas DataFrame.

Parameters

Parameter	Type	Default	Description
data	pd.DataFrame	required	The DataFrame to explore
options	ViewdtOptions	viewdt_options()	Configuration object — see viewdt_options() below
dataset_name	str	inferred	Variable name substituted in generated code. Auto-detected from the call site if omitted

Returns ViewdtWidget

Widget methods

Method	Description
.show()	Open the widget in the default web browser
.save(path)	Export to a standalone HTML file
_repr_html_()	Renders inline in Jupyter / VS Code notebooks automatically

---

viewdt_options(**kwargs) → ViewdtOptions

Build a configuration object to pass to viewdt().
All parameters are keyword-only and have sensible defaults — pass only what you want to change.

from viewdt import viewdt, viewdt_options

viewdt(df, options=viewdt_options(
    theme="dark",
    hidden_columns=["internal_id", "created_at"],
    hist_bins=30,
    top_n=15,
))

Options reference

Parameter	Type	Default	Description
theme	"auto" | "light" | "dark"	"auto"	UI colour scheme. "auto" follows the OS preference
show_labels	bool	True	Show variable labels from series.attrs["label"]
histograms	bool	True	Render spark-histograms / category bars in column headers
missing_bars	bool	True	Show the data-completeness bar at the bottom of each header
type_badges	bool	True	Show data-type badges in column headers
insights	bool	True	Enable the Data Insights drawer (click any column header)
query_builder	bool	True	Enable the visual query builder
column_picker	bool	True	Enable the column-visibility picker
code_export	bool	True	Enable the reproducible code generator
global_search	bool	True	Enable the global search box
na_string	str	"NA"	Text displayed for missing values
hidden_columns	list[str] | None	None	Columns hidden on initial render
page_size	int	200	Rows kept in the virtualised DOM buffer
hist_bins	int	20	Bin count for numeric / datetime histograms
top_n	int	10	Number of categories profiled for text columns
max_cells	int	5_000_000	Warn when nrow × ncol exceeds this threshold

---

save_viewdt(data, path, options=None, dataset_name=None)

Convenience wrapper — profiles data and writes the widget directly to path.

from viewdt import save_viewdt

save_viewdt(df, "report.html")
save_viewdt(df, "report_dark.html", options=viewdt_options(theme="dark"))

---

Built-in datasets

Seven sample datasets are included — no internet connection required.

Loader	Rows	Columns	Description
load_iris()	150	5	Fisher's Iris flower measurements — setosa, versicolor, virginica
load_mtcars()	32	12	Motor Trend Car Road Tests 1974 — mpg, hp, weight, transmission
load_penguins()	344	8	Palmer Archipelago penguins — bill & flipper measurements (with NaNs)
load_tips()	244	7	Restaurant tips — bill, tip, sex, smoker, day, time, party size
load_gapminder()	444	6	Gapminder — country, continent, year, life expectancy, population, GDP
load_titanic()	891	9	Titanic passengers — survival, class, sex, age (with NaNs), fare
load_stocks()	1260	4	Daily closing prices — AAPL, GOOG, MSFT, AMZN, META (datetime column)

from viewdt import (
    viewdt, viewdt_options,
    load_iris, load_mtcars, load_penguins,
    load_tips, load_gapminder, load_titanic, load_stocks,
    list_datasets,
)

# See all available datasets
viewdt(list_datasets())

# Iris — type badges, spark histograms, column labels
viewdt(load_iris())

# mtcars — all-numeric, code export
viewdt(load_mtcars(), options=viewdt_options(theme="dark"))

# Penguins — completeness bars (missing values in age, sex)
viewdt(load_penguins())

# Gapminder — try the query builder: continent = "Asia", year >= 1990
viewdt(load_gapminder())

# Titanic — filter survived = 1, pclass = 1 to explore first-class survivors
viewdt(load_titanic(), options=viewdt_options(hidden_columns=["name"]))

# Stocks — datetime column, time-series data
viewdt(load_stocks())

---

Examples

Dark theme with pre-hidden columns

from viewdt import viewdt, viewdt_options

viewdt(
    df,
    options=viewdt_options(
        theme="dark",
        hidden_columns=["row_id", "updated_at"],
    ),
)

Clinical / labelled data (haven / ADaM)

Column labels stored in series.attrs["label"] are shown below the column name in the header — the same behaviour as the R ViewR package with haven-imported datasets.

df["AVAL"].attrs["label"] = "Analysis Value"
df["PARAMCD"].attrs["label"] = "Parameter Code"

viewdt(df)  # labels appear in every column header

Lightweight view — disable heavy features

viewdt(
    df,
    options=viewdt_options(
        histograms=False,
        insights=False,
        query_builder=False,
    ),
)

Large DataFrames

# viewdt warns automatically when nrow × ncol > max_cells (default 5 M).
# Sample before exploring if needed:
viewdt(df.sample(50_000))

# Or raise the threshold:
viewdt(df, options=viewdt_options(max_cells=20_000_000))

Export to HTML for sharing

from viewdt import save_viewdt, viewdt_options

save_viewdt(
    df,
    "team_report.html",
    options=viewdt_options(theme="light", code_export=False),
    dataset_name="sales_q1",
)

---

Query builder operators

The visual query builder exposes type-appropriate operators for each column:

Column type	Available operators
Numeric	= ≠ < ≤ > ≥ is null not null
Text	= ≠ contains !contains in !in is null not null
Logical	is true is false is null not null
Datetime	= < ≤ > ≥ is null not null

The in / !in operators accept a comma-separated list of values in the input field.

---

Code export

The Code button generates reproducible code that matches the current filter state and column selection in three dialects:

pandas

mask = (
    df['category'].isin(['A', 'B']) &
    (df['price'] > 50)
)
df = df[mask]
df = df[['category', 'price', 'score']]

Python

import pandas as pd

result = df[
    df['category'].isin(['A', 'B']) &
    (df['price'] > 50)
]
result = result[['category', 'price', 'score']]

SQL

SELECT "category", "price", "score"
FROM "df"
WHERE "category" IN ('A', 'B')
  AND "price" > 50;

---

Comparison with R ViewR

Feature	R viewdt()	Python viewdt()
Data input	data.frame / tibble	pd.DataFrame
Column profiling	In R, before render	In Python, before render
Spark charts	SVG, vanilla JS	SVG, vanilla JS
Code export	dplyr / base R / SQL	pandas / Python / SQL
Column labels	label attribute (haven)	series.attrs["label"]
Standalone output	htmlwidget	ViewdtWidget → .save()
Jupyter support	via htmlwidgets	via _repr_html_()
Dependencies	R packages	pandas, numpy
JS dependencies	None	None

---

Development

git clone https://github.com/itsmdivakaran/viewdt-python
cd viewdt-python
pip install -e ".[notebook]"

Run the smoke test:

python -c "
import pandas as pd, numpy as np
from viewdt import viewdt
df = pd.DataFrame({'x': np.random.randn(100), 'y': list('ABCD') * 25})
viewdt(df).show()
"

---

License

MIT License — see LICENSE for details.

---

Author

Mahesh Divakaran
Research Scholar, Amity University Uttar Pradesh

• GitHub: github.com/itsmdivakaran
• ORCID: 0000-0002-3488-0857
• Email: imaheshdivakaran@gmail.com

---

Citation

If you use viewdt in academic work, please cite the original R package:

Divakaran M (2026). ViewR: Interactive Data Viewer, Filter, and Editor. R package version 0.2.0. https://itsmdivakaran.github.io/viewR/

BibTeX:

@Manual{ViewR,
  title  = {{ViewR}: Interactive Data Viewer, Filter, and Editor},
  author = {Mahesh Divakaran},
  year   = {2026},
  note   = {R package version 0.2.0},
  url    = {https://itsmdivakaran.github.io/viewR/},
}

---

Acknowledgements

Python port of the ViewR R package (CRAN). The widget design, feature set, and viewdt_options() API are derived directly from the R implementation by the same author.