edazer 0.1.4

lightweight library that provides functionalities for common EDA tasks

Edazer

Edazer is a lightweight Python package for performing common exploratory data analysis (EDA) tasks. It provides quick and intuitive methods to inspect, summarize, and understand datasets—supporting both pandas and polars backends.

Includes utilities for: Interactive DataFrame exploration (via itables)

Automated profiling reports (via a wrapper around ydata-profiling)

Unique key detection (via get_primary_key)

🚀 Ideal for:

Jupyter notebooks

Fast, one-line data profiling

Early-stage dataset exploration

---

Features

• Quick DataFrame Summaries: Instantly view info, describe, nulls, duplicates, and shape using summary method
• Unique Value Inspection: Easily display unique values for any or all columns.
• Type-based Column Selection: Find columns by dtype (e.g., int, float categorical).
• Flexible Subsetting: Use the lookup method to view head, tail, or random samples.
• Custom DataFrame Naming: Track multiple DataFrames with custom names for clarity.
• Primary Key Detection: Automatically identify single or multi-column combinations that can serve as unique identifiers.

---

Installation

pip install edazer

---

Quick Start with Titanic Dataset

import seaborn as sns
from edazer import Edazer, interactive_df 
from edazer.profiling import show_data_profile

# Enable interactive DataFrames (via itables)
interactive_df()

# Load dataset
titanic = sns.load_dataset('titanic')

# Initialize Edazer instance
titanic_dz = Edazer(titanic, backend="pandas", name="titanic")

# Complete DataFrame summary
titanic_dz.summarize_df()

# Data profiling report (via ydata_profiling)
show_data_profile(titanic_dz)

# Show unique values for specific columns
titanic_dz.show_unique_values(column_names=['class', 'embarked'], max_unique=5)

# Get float columns
print(titanic_dz.cols_with_dtype(['float'], exact=False))

# Combine methods: get object columns and show their unique values
titanic_dz.show_unique_values(column_names=titanic_dz.cols_with_dtype(dtypes=["object"]))

# View first few rows
print(titanic_dz.lookup("head"))

# Access raw DataFrame
print(titanic_dz.df.columns)

---

📘 API Reference

Edazer(df, backend="pandas", name=None)

Create an analyzer instance.

• df: pd.DataFrame or pl.DataFrame
• backend: "pandas" or "polars" (default: "pandas")
• name: Optional string label for the DataFrame

---

summarize_df()

Print summary:

• Schema/info
• Descriptive stats
• Null/duplicate counts
• Unique values
• Shape

---

show_unique_values(column_names=None, max_unique=10)

Show unique values for columns.

• column_names: Optional list of columns
• max_unique: Max unique values to display per column

---

cols_with_dtype(dtypes, exact=False, return_dtype_map=False)

Return columns matching specified dtypes.

• dtypes: List of type strings (e.g. ["int", "object"])
• exact: Match full dtype string (e.g. "int64")
• return_dtype_map: If True, return {col: dtype}

---

lookup(option="head")

Quickly inspect data.

• option: "head", "tail", or "sample"

---

🆕 get_primary_key(df, threshold=0.9, n_combos=1, valid_column_dtypes=None)

Identify column(s) or column combinations that can serve as unique keys.

Parameters

• df – The input DataFrame.

• threshold – Proportion of uniqueness required (default = 0.9).

• n_combos– Number of columns to combine when testing composite keys (default = 1).

• valid_column_dtypes – Data types to consider (default = ["int", "datetime64", "object"]).

Returns

List[str] or List[List[str]]: Candidate key columns or combinations that are likely unique identifiers.

Example usage

from edazer import get_primary_key
import pandas as pd

df = pd.DataFrame({
    "id": [1, 2, 3, 4],
    "name": ["A", "B", "C", "A"],
    "date": pd.date_range("2023-01-01", periods=4)
})

get_primary_key(df, threshold=1.0, n_combos=2)
# Output: [['id', 'name'], ['id', 'date'], ['name', 'date']]

Example Output

titanic_eda.show_unique_values(column_names=titanic_dz.cols_with_dtype(dtypes=["object"]))

# Output:
sex: ['male', 'female']
embarked: ['S', 'C', 'Q', nan]
who: ['man', 'woman', 'child']
embark_town: ['Southampton', 'Cherbourg', 'Queenstown', nan]
alive: ['no', 'yes']

---

Contributing

Contributions are highly welcome!

https://github.com/adarsh-79/edazer

---

License

MIT License

---

Author

adarsh3690704