Data profiling for pandas DataFrames.
The released version number is still 1.1.5. The working tree is the Pytics 2.0 analytical system: semantic inference, dataset analysis, and a structured result. A notebook display of that result is a compact HTML view. Interactive HTML and PDF are not current export features.
pip install pyticsRuntime dependencies are pandas, NumPy, SciPy, and scikit-learn. Plotly, Jinja2, xhtml2pdf, ReportLab, Kaleido, and Matplotlib are not installed with Pytics.
Load a DataFrame yourself, then profile it:
import pandas as pd
import pytics
df = pd.read_csv("your_data.csv")
report = pytics.profile(df)
report
comparison = pytics.compare(df, df)report is a ProfileResult. comparison is a ComparisonResult. In Jupyter or Colab, displaying either object uses _repr_html_(). That view reads the result. It does not recompute the analysis, and it does not write a file.
profile_result and comparison_result remain in pytics.results. The top-level functions call them. A keyword-only target is passed through when one is supplied. File export arguments from the 1.1.5 renderer are not accepted.
The configuration object and semantic overrides are still open, recorded as OPEN-004 in docs/pytics-2/DECISIONS.md.
Pytics 1.1.5 generated an HTML report, and it attempted a PDF report, from profile() and compare(). That renderer analyzed the DataFrame on its own, with Plotly charts, Jinja templates, and xhtml2pdf. It is not the 2.0 analysis. It has been removed, including its templates and its PDF test. There is no Markdown report exporter.
The durable 2.0 design notes are in docs/pytics-2/. They are documentation, not a report format.
Contributions are welcome. See CONTRIBUTING.md.
This project is licensed under the MIT License. See the LICENSE file for details.