Mid Year, New overviewpy? 🎉

It’s been a while since the first release of overviewpy, and I’m excited to share what’s new in v0.2.0 — it’s a big one! The package has grown substantially from its humble beginnings as a table-generating features into a full(er) toolkit for exploratory data analysis of panel data. Let me walk you through what’s new.

A New API: The Overview Class

The biggest structural change in v0.2.0 is that overviewpy now has a proper object-oriented interface. Instead of calling standalone functions, you create an Overview object once and then call methods on it:

from overviewpy.overview import Overview
import pandas as pd

data = {
    'id': ['RWA', 'RWA', 'RWA', 'GAB', 'GAB', 'FRA'],
    'year': [2022, 2023, 2021, 2023, 2020, 2019]
}

df = pd.DataFrame(data)
overview = Overview(df=df, id_col='id', time='year')

This makes for cleaner, more readable code — you define your data and columns once, and every function call flows naturally from there. A big thank you to Barrett Smith for bringing this idea to life — the OOP approach was his initiative, and it introduced a whole new approach of how you can use overviewpy. And speaking of those calls, there are a lot of new ones!

Method Calls in overviewpy

overview_tab — Automatic NA Handling

overview_tab hasn’t changed in what it produces — a compact two-column summary of which ids appear in which time periods — but it now handles missing values for you automatically. Rows with NAs in id_col or time are silently dropped with a UserWarning, so no extra preprocessing needed.

df_overview = overview.overview_tab()

overview_summary — Know Your Data at a Glance

New in v0.2.0 is overview_summary, which returns a per-column summary of any data frame: non-null count, number of unique values, and a few sample values. This feature was also introduced by Barrett Smith - thank you so much for adding it!

overview.overview_summary()

Think of it as a friendly first look at any unfamiliar dataset.

overview_na — Missing Values at a Glance

overview_na plots an overview of missing values across your dataset. New in v0.2.0 are several extra parameters:

• you can now switch between percentages and absolute counts,
• flip to a row-wise view,
• and control the y-axis scale.

overview.overview_na(perc=True, row_wise=False)

overview_plot — Visualize Observation Coverage

Ever wanted to see at a glance which units are observed in which time periods? overview_plot generates a connected dot-plot that maps each id against its time coverage:

overview.overview_plot()

We perceived it as valuable in overviewR and hopefully it also serves as an intuitive way to spot gaps or irregular coverage in the data of the Python community.

overview_heat — A Heatmap of Your Sample

For a more granular take on coverage, overview_heat produces a heatmap of observation counts (or percentages) for each id–time combination. The darker the cell, the more observations you have — making sparse periods immediately visible:

overview.overview_heat()

overview_overlap — Compare Two Data Frames

How much do two datasets actually overlap? overview_overlap answers that question visually, with either a bar chart or a Venn diagram comparing the id coverage of two data frames:

overview.overview_overlap(df2=other_df, id_col2='country')

overview.overview_overlap(df2=other_df, id_col2='country', plot_type='venn')

This is especially useful when merging datasets and you want to know what you’d be losing.

overview_crossplot & overview_crosstab — Split by Conditions

Sometimes you want to examine how observations are distributed across two conditions simultaneously. overview_crossplot creates a scatter plot divided into four quadrants by two user-defined thresholds, while overview_crosstab produces a 2×2 cross-table of the same:

overview.overview_crossplot(var1='gdp', var2='pop', threshold1=25000, threshold2=2.75e7)

overview_markdown & overview_latex — Export-Ready Output

Once you have your overview_tab summary, you can now export it directly to Markdown or LaTeX — no manual formatting needed.

print(overview.overview_markdown())

## Time and scope of the sample

| Sample | Time frame |
|---|---|
| ARG | 2002 |
| BEL | 2013-2014 |
| FRA | 2015, 2019 |
| GAB | 2020, 2023 |
| RWA | 2021-2023 |

overview_markdown is great for dropping tables straight into a blog post or README, while overview_latex produces publication-ready output you can paste into a paper:

overview.overview_latex(save_out=True, file_path='sample_table.tex')

Getting Started

Install the latest version from PyPI:

pip install overviewpy

For the full API reference and worked examples, check out the documentation.

Did you get a chance to give it a try? I’d love to hear your feedback — feel free to open an issue or start a discussion on GitHub!