ezclean-data 0.1.0

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

✨ ezclean-data

A premium, dataset-agnostic Python library designed to automate the painful parts of data loading, cleaning, and exploration. With ezclean-data, you can load any structured file format, sanitize outliers and null values, and instantly produce beautiful, interactive visualization dashboards or multi-variable pairplot matrices.

---

🚀 Features

• 📥 Smart Data Loader: Auto-detects extensions (CSV, Excel, Parquet, JSON, etc.) and routes them to optimized Pandas engines, streaming directly over HTTP or loading local paths.
• 🧼 Intelligent Data Cleaner: Standardizes columns to snake_case, handles structural garbage strings, handles outliers via IQR boundary thresholds, and fills null values using type-specific heuristics (e.g. median for numbers).
• 📊 Universal Plot Grid: Renders a generalized interactive Plotly pairplot matrix of subplots showing all possible univariate distributions (diagonal) and bivariate relationships (off-diagonal) for any dataset.
• 🎨 Standalone HTML Dashboard: Generates a fully interactive, lightweight dashboard with statistics cards, a column definitions table, and a dynamic JavaScript plot builder that works fully offline!

---

📦 Installation

Install ezclean-data directly from PyPI:

pip install ezclean-data

---

⚡ Quick Start

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

# 1. Load your dataset from a file or url
df = Smart_loader("tested.csv")

# 2. Run the unified cleaning pipeline
df_cleaned = Cleaner(df)

# 3. Print column statistics
colname(df_cleaned)

# 4. Plot a single column (auto-detects types)
plot(df_cleaned, "survived")

# 5. Plot the generalized pairplot matrix (all combinations)
plot(df_cleaned)

# 6. Generate and open a gorgeous standalone HTML Dashboard
plot_dashboard(df_cleaned, filename="my_dashboard.html")

---

🛠️ Module API Overview

1. Smart_loader(file_path, **kwargs)

Instantly routes local or remote URLs to Pandas readers. Supported formats: csv, tsv, txt, json, jsonl, ndjson, excel (xlsx, xls, ods), parquet, feather, arrow, orc, xml, html, pickle, stata, spss, sas, hdf.

2. Cleaner(df, ...)

High-level cleaning pipeline wrapping:

• column_name_sanity(): Clean symbols, Deduplicate underscores, Convert to snake_case.
• sanitize_data(): Replaces structural garbage (?, NULL, nil) with NumPy NaNs.
• text_normalization(): Trims whitespace and normalizes string fields.
• auto_type_correction(): Converts column dtypes to numeric if >50% of values match.
• intelligent_null_filling(): Median imputes numeric fields; fills categorical values with "Unknown".
• handle_outliers(): IQR-based outlier trimming.

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

• If target_column is provided, renders a single visual (numerical gets Box+Histogram; categorical gets Donut+Bar; datetime gets Line Trend).
• If target_column=None, renders plot_matrix containing all univariate and bivariate subplots for selected columns (default: top 5).

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

Writes a self-contained, interactive HTML dashboard containing:

1. Complete column completeness summary tables.
2. Dynamic Plotly client visualizer where users can build custom X vs Y plots.
3. Embedded pairplot relation matrix.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.