ezclean 0.1.0

A robust, dataset-agnostic loader, cleaner, and automated interactive visual pairplot dashboard engine.

ezclean-data

A comprehensive, dataset-agnostic Python library for automated data ingestion, cleaning, preprocessing, and exploratory data analysis.

ezclean-data streamlines the repetitive tasks involved in preparing datasets by providing intelligent loading, automated sanitization, statistical summaries, advanced visualizations, and standalone interactive dashboards.

---

Overview

Data preparation often consumes a significant portion of the data science workflow. ezclean-data provides a unified interface that automatically loads structured datasets, cleans inconsistencies, handles missing values and outliers, standardizes column names, and generates insightful visualizations with minimal code.

The library is designed to work across a wide variety of dataset formats and structures, making it suitable for students, researchers, analysts, and machine learning practitioners.

---

Features

Smart Data Loading

• Automatically detects file formats and selects the appropriate Pandas engine.
• Supports both local files and remote URLs.
• Handles a broad range of structured data formats.

Automated Data Cleaning

• Standardizes column names into consistent snake_case format.
• Detects and replaces common invalid placeholder values.
• Performs automatic type correction where appropriate.
• Handles missing values using intelligent, data-type-aware strategies.
• Detects and mitigates outliers using the Interquartile Range (IQR) method.

Exploratory Data Analysis

• Generates detailed column summaries and completeness statistics.
• Provides automatic visualizations based on column data types.
• Creates generalized pairplot-style relationship matrices for rapid exploration.

Interactive Dashboard Generation

• Produces self-contained HTML dashboards.
• Includes summary statistics and data quality metrics.
• Provides interactive Plotly-based visualizations.
• Works entirely offline once generated.

---

Installation

Install directly from PyPI:

pip install ezclean-data

---

Quick Start

from ezclean import Smart_loader, Cleaner, colname, plot, plot_dashboard

# Load dataset
df = Smart_loader("tested.csv")

# Execute cleaning pipeline
df_cleaned = Cleaner(df)

# Display column statistics
colname(df_cleaned)

# Visualize a single column
plot(df_cleaned, "survived")

# Generate a relationship matrix
plot(df_cleaned)

# Create an interactive dashboard
plot_dashboard(
    df_cleaned,
    filename="my_dashboard.html"
)

---

API Reference

Smart_loader()

Smart_loader(file_path, **kwargs)

Automatically loads structured datasets from local storage or remote URLs.

Supported Formats

Category	Formats
Text Files	CSV, TSV, TXT
JSON Formats	JSON, JSONL, NDJSON
Spreadsheet Files	XLSX, XLS, ODS
Columnar Formats	Parquet, Feather, Arrow, ORC
Statistical Formats	SPSS, SAS, Stata
Other Formats	XML, HTML, Pickle, HDF

---

Cleaner()

Cleaner(df, ...)

Executes a complete data-cleaning pipeline.

Included Operations

Column Name Standardization

• Converts names to snake_case
• Removes special characters
• Eliminates duplicate separators

Data Sanitization

Replaces common placeholder values such as:

?
NULL
null
nil
N/A
NaN

with proper missing-value representations.

Text Normalization

• Trims whitespace
• Standardizes string formatting

Automatic Type Detection

• Converts columns to numeric types when appropriate
• Preserves incompatible values

Missing Value Handling

• Numerical columns → Median Imputation
• Categorical columns → "Unknown" Replacement

Outlier Treatment

• Uses Interquartile Range (IQR) thresholds
• Removes extreme observations automatically

---

colname()

colname(df)

Displays detailed metadata for each column, including:

• Data type
• Missing value count
• Completeness percentage
• Unique value count
• Statistical summaries

---

plot()

plot(df, target_column=None, columns=None)

Single-Column Visualization

When a target column is specified, the visualization is selected automatically based on data type.

Data Type	Visualization
Numeric	Histogram + Box Plot
Categorical	Bar Chart + Donut Chart
Datetime	Trend Line

Relationship Matrix

plot(df)

Generates a generalized pairplot matrix displaying:

• Univariate distributions
• Correlation patterns
• Relationships between variables

---

plot_dashboard()

plot_dashboard(
    df,
    filename="ezclean_dashboard.html",
    show=True
)

Creates a standalone interactive dashboard containing:

Dataset Summary

• Dataset dimensions
• Completeness metrics
• Data quality statistics

Column Analysis

• Data types
• Missing values
• Unique value counts

Interactive Visualization Builder

Users can dynamically select:

• X-axis variables
• Y-axis variables
• Plot types

without writing additional code.

Relationship Matrix

Embedded Plotly-based exploratory visualization for multivariate analysis.

---

Example Workflow

from ezclean import *

df = Smart_loader("data.csv")

df = Cleaner(df)

colname(df)

plot(df, "age")

plot(df)

plot_dashboard(
    df,
    filename="dashboard.html"
)

---

Use Cases

• Data Science Projects
• Machine Learning Preprocessing
• Academic Research
• Exploratory Data Analysis
• Business Intelligence Reporting
• Rapid Dataset Validation
• Educational Applications

---

Why ezclean-data?

Most data analysis projects begin with repetitive preprocessing tasks such as loading files, cleaning columns, handling missing values, detecting outliers, and creating exploratory visualizations.

ezclean-data consolidates these operations into a simple and consistent workflow, allowing users to focus on analysis and model development rather than boilerplate data preparation code.

---

License

This project is licensed under the MIT License.

See the LICENSE file for complete licensing information.

---

Author

Developed and maintained by Thilac Ramesh.

Contributions, feature requests, and issue reports are welcome.