📊 dfglimpse

A quick, per-column glimpse of a pandas DataFrame for exploratory data analysis (EDA).

glimpse summarizes each column of a DataFrame in a single table — its dtype, how many values are missing (count and percentage), how many unique values it has, and either the full set of unique values or a short preview for high-cardinality columns.

Installation

pip install dfglimpse

Quick start

import pandas as pd
from dfglimpse import glimpse

df = pd.read_csv("sample_data.csv")
glimpse(df)

Output:

       Column  dtypes  NaN Count  NaN %  Nunique                               Unique Values
           Id   int64          0    0.0       20                    [1, 2, 3] ... (+17 more)
   MSSubClass   int64          0    0.0        5                        [20, 60, 70, 50, 30]
     MSZoning  object          0    0.0        4                       [RL, RM, C (all), RH]
  LotFrontage float64          4   20.0       15           [65.0, 80.0, 70.0] ... (+12 more)
      LotArea   int64          0    0.0       20          [8450, 9600, 11250] ... (+17 more)
SaleCondition  object          0    0.0        5 [Normal, Abnorml, Partial, AdjLand, Alloca]

The result is an ordinary DataFrame, so you can sort or filter it — for example, glimpse(df).sort_values("NaN %", ascending=False) to find the columns with the most missing values.

API

glimpse(df, columns=None, max_unique=10, n_samples=3) -> pandas.DataFrame

Returns a DataFrame with one row per column and the columns Column, dtypes, NaN Count, NaN %, Nunique, Unique Values.

Missing values are excluded from Nunique and from the displayed unique values, so they stay consistent with each other.

Requirements

• Python 3.10+
• pandas 2.0+

License

MIT License. See LICENSE.

日本語

dfglimpse は、pandas の DataFrame の各列を一望するための小さな EDA 補助ツールです。 glimpse(df) を呼ぶと、各列について以下を1つの表にまとめて返します。

• データ型（dtype）
• 欠損（NaN）数と割合（%）
• ユニークな値の個数（欠損は除外）
• ユニークな値の内容（多い列は先頭数件のプレビュー）

インストール

pip install dfglimpse

使い方

import pandas as pd
from dfglimpse import glimpse

df = pd.read_csv("sample_data.csv")
glimpse(df)

戻り値はただの DataFrame なので、glimpse(df).sort_values("NaN %", ascending=False) のように並べ替え・絞り込みができます。

引数