Overview

EazyDataFix 0.3.0 introduces a complete deterministic Agentic EDA workflow. Every step is reproducible, traceable and runs entirely without an LLM.

Highlights

deterministic exploratory data analysis
semantic column-role detection
deterministic planning and execution
traceable findings and follow-up actions
visualisation recommendations
unresolved domain questions
standalone HTML reports
stable JSON reports
optional Markdown reports
deterministic PNG visualisations
shared dataset validation
DataFrame non-mutation
Python 3.10–3.13 support

Quick start

agentic_eda.py

python

Colab GitHub

import eazydatafix as edf

workflow = edf.run_agentic_eda("employees.csv")

report = edf.export_agentic_eda_report(
    workflow,
    dataset="employees.csv",
    output_dir="eda-report",
)

print(workflow.deterministic_final_summary)
print(report.generated_files)

Generated output

eda-report/

text

eda-report/
├── agentic-eda-report.html
├── agentic-eda-report.json
├── agentic-eda-report.md
└── visualisations/
    ├── 01-missing-value-chart-phone-salary.png
    ├── 02-bar-chart-department.png
    └── 03-time-series-line-chart-joining-date.png

Deterministic EDA

edf.eda(...) runs a deterministic exploratory data analysis with semantic role detection — no hidden model behaviour, no random branching.

EDA Planner

edf.plan_eda(...) selects or skips relevant EDA steps with priorities, dependencies and clear reasons.

EDA Executor

edf.execute_eda(...) runs modular deterministic analyses for missing values, duplicates, distributions, outliers, skewness, imbalance, correlations and datetime trends. Each step is isolated so a single failure does not halt the pipeline.

Agentic EDA Orchestrator

edf.run_agentic_eda(...) coordinates understanding, planning, execution and follow-up decisions in one call. Every finding includes its source step, priority, target columns, reason and prerequisites.

Report Export

edf.export_agentic_eda_report(...) writes standalone HTML, stable JSON, optional Markdown and deterministic PNG visualisations to a target directory.

Public API added in 0.3.0

edf.eda(...)

Generate a structured deterministic EDA result.

edf.plan_eda(...)

Create a reproducible follow-up analysis plan.

edf.execute_eda(...)

Execute selected EDA steps using deterministic handlers.

edf.run_agentic_eda(...)

Run the complete deterministic Agentic EDA workflow.

edf.export_agentic_eda_report(...)

Export HTML, JSON, Markdown, and supported PNG visualisations.

Compatibility

Python 3.10, 3.11, 3.12 and 3.13. Caller-owned pandas DataFrames are copied and preserved.