ResidentMario/missingno
Missingno
missingno es una biblioteca de Python para la visualización y el análisis de patrones de datos faltantes. Proporciona un conjunto de herramientas para perfilar la integridad de los conjuntos de datos, mapear brechas de datos y cuantificar el volumen de valores nulos en todas las variables.
La biblioteca se diferencia por un analizador de correlación de nulidad y una herramienta de clustering jerárquico de datos. Estos componentes permiten la detección de dependencias y tendencias sistémicas midiendo cómo la ausencia de una variable se relaciona con la ausencia de otra.
El conjunto de herramientas cubre capacidades más amplias de auditoría de calidad de datos y análisis exploratorio. Incluye funciones para el resumen de nulidad de columnas utilizando escalas lineales y logarítmicas, así como mapeo basado en matrices para identificar brechas sistémicas en los registros.
Features
- Missing Data Analysis - Identifies patterns and systemic gaps in datasets using Python to determine how and where information is missing.
- Hierarchical Clustering - Implements hierarchical clustering to group variables based on similar missingness patterns using dendrograms.
- Feature Correlation Analysis - Measures statistical relationships between the absence of one variable and another using correlation heatmaps.
- Data Analysis & Visualization - Provides tools for the visual analysis and mapping of data completion patterns to identify systemic gaps.
- Dataframe Visualizers - Provides a pipeline that transforms tabular pandas dataframes into static visual representations for missing data exploration.
- Exploratory Data Analysis - Enables exploratory data analysis by visualizing the distribution and volume of null values.
- Missing Data Clustering - Groups variables using hierarchical clustering to reveal deep trends and dependencies in how data is missing across a dataset.
- Completeness Profilers - Provides a visual summary of missing value volumes per column using linear and logarithmic scaling.
- Completeness Summaries - Displays the volume of missing values for each variable using linear or logarithmic scales to compare completeness.
- Correlation Matrices - Generates correlation matrices to compute the statistical relationship between missingness in different variables.
- Nullity Correlation Analyzers - Provides a visualization tool that measures how the absence of one variable relates to the absence of another.
- Nullity Masks - Converts dataframes into binary masks of presence and absence to identify systemic gaps through visual patterns.
- Dataset Quality Analysis - Analyzes nullity correlations and dependencies between variables to detect biases or errors in data collection.
- Data Preprocessing for Modeling - Prepares datasets for machine learning by analyzing missingness patterns to inform imputation or removal strategies.
- Logarithmic Axis Scales - Supports logarithmic axis scales to visualize and compare null counts across variables with vastly different volumes.
- Automated EDA and Visualization - Visualize missing data patterns.
- Data Processing Libraries - Visualizing and diagnosing missing data patterns.
- Data Visualization - Visualization module for missing data patterns.
- Python Visualization - Visual utility for assessing dataset completeness.
- Python Visualization Libraries - Visualization utilities for checking dataset completeness.
