nextgen-v0.1.0-alpha

Security notice

This Alpha release is unsigned unless its assets display a trusted Windows code-signing certificate. Windows may show an unknown-publisher warning. Verify SHA256SUMS.txt before use; this prerelease is not intended for production deployment.

Full Changelog: v2.3.0...nextgen-v0.1.0-alpha

DataSense v2.3.0

DataSense 2.3.0 — Verified Local Exports

DataSense 2.3.0 strengthens the Windows desktop release with a verified export workflow for sensitive analytical work. Analysts can now create a styled HTML report or interactive dashboard only after the active Trust Center evidence passes its local quality and schema gates. The workflow produces a companion, signed receipt that binds the permitted action to current metadata-only evidence.

Downloads

File	Use it for
DataSense-2.3.0-setup.exe	Recommended Windows installer with Start-menu shortcut, optional desktop shortcut, and uninstaller.
DataSense-2.3.0-windows-x64-portable.zip	Portable bundle; unzip and run DataSense.exe without installation.
DataSense.exe	Standalone executable from the packaged bundle for controlled deployment tooling.

Windows 10 or Windows 11, 64-bit, is required. Python is bundled and is not required on the user’s computer.

Verified export workflow

Choose File → Export verified artifact to select either an analysis report or an interactive dashboard. The workflow requires current Trust Center checks and a local HMAC signing key. DataSense evaluates the requested internal artifact against the signed evidence and only writes the HTML artifact when the decision is allow.

Output	Behaviour
Verified HTML artifact	Written only after quality and schema gates permit the requested action.
*.trust-receipt.json companion file	Signed, metadata-only decision evidence stored next to the requested artifact.
Blocked or approval-required action	The HTML artifact is not written; a metadata-only receipt records the decision and reason codes for offline review.

The signing key is never included in the receipt and must remain outside source control. The receipt does not include raw dataset values, local source paths, recipient details, URLs, prompts, or credentials.

Reliability and packaging

The Windows build and release workflows now install both desktop and Enterprise Control Plane test dependencies before invoking the complete regression suite. This removes the prior gap in which the packaging workflows collected Control Plane tests without installing their declared dependencies.

DataSense 2.3.0 retains its full desktop studio for data import, profiling, preparation, statistics, visualization, SQL analysis, time series, machine learning, AutoML, dashboards, local project files, and Trust Center evidence.

Upgrade note

Open existing .dsproj projects normally. Existing data and analysis workflows are preserved. To use verified export, create or review a Trust Center contract, run the current checks, and select a local signing key when prompted.

Full Changelog: v2.2.1...v2.3.0

DataSense v2.2.1

DataSense 2.2.1 — Windows Packaging Hotfix

DataSense 2.2.1 is the production packaging patch for the Trusted Analytics release. It preserves the Trust Center, data contracts, audit evidence and desktop stability repairs introduced in 2.2.0, while correcting the Windows pipeline so it installs and invokes the regression runner deterministically before packaging.

Downloads

File	Use it for
DataSense-2.2.1-setup.exe	Recommended installer with Start-menu shortcut, optional desktop shortcut and uninstaller.
DataSense-2.2.1-windows-x64-portable.zip	Portable bundle; unzip it and run DataSense.exe without installation.
DataSense.exe	Standalone executable from the packaged bundle for controlled deployment tooling.

Windows 10 or Windows 11, 64-bit, is required. Python is bundled and is not required on the user’s computer.

Trust Center

Trust Center does not transmit the active dataset. It provides a local sensitive-data scan that identifies likely email addresses, telephone numbers, IP addresses and payment-card patterns without retaining detected values. Findings are signals for review, not a legal classification or automated access-control decision.

Data contracts let an analyst define repeatable acceptance criteria. The initial rule set supports populated-value, uniqueness, numeric range, allowed-value, regular-expression and timestamp-freshness controls. Every rule has an explicit severity, and recommended rules are visible and editable before use. Running a check produces a weighted quality score together with Trusted, Needs attention, Blocked, or Not configured status. A data mutation invalidates the prior result automatically so stale evidence is never presented as current.

Output	Purpose
Trust Center table	Review each control, observed condition, expectation, violations and error state.
JSON audit evidence	Attach deterministic, portable validation evidence to an approval process, ticket or archive.
Styled HTML analysis report	Includes the current Trust Center summary and control results when checks were run.
.dsproj project	Preserves the contract definition; DataSense intentionally asks the user to rerun validation after restoring a project.

Included platform capabilities

DataSense remains a full desktop studio for data import, quality profiling, preparation, statistics, visualisation, SQL analysis, time series, machine learning, AutoML, dashboards, local project files and styled reports. The release pipeline builds the Windows x64 bundle, portable archive and installer from the tagged source revision; it also runs regression tests before publishing release assets.

Validation performed

The release candidate passed the complete automated suite, including dedicated contract, sensitive-data scanning, project persistence and off-screen desktop startup coverage. The PyInstaller recipe was additionally exercised in a Linux validation environment; Windows release artifacts are built only by the dedicated windows-latest GitHub Actions pipeline.

Upgrade note

Open existing projects normally. Their datasets and earlier analysis features are preserved. Existing projects do not contain a data contract until one is created in Trust Center.

Full Changelog: v2.2.0...v2.2.1

DataSense v2.2.0

DataSense 2.2.0 — Trusted Analytics

DataSense 2.2.0 adds Trust Center, a local-first governance workspace that helps analysts establish whether a dataset is fit for reporting, modelling or sharing. This release also repairs project persistence and workspace-refresh defects found during desktop startup validation.

Downloads

File	Use it for
DataSense-2.2.0-setup.exe	Recommended installer with Start-menu shortcut, optional desktop shortcut and uninstaller.
DataSense-2.2.0-windows-x64-portable.zip	Portable bundle; unzip it and run DataSense.exe without installation.
DataSense.exe	Standalone executable from the packaged bundle for controlled deployment tooling.

Windows 10 or Windows 11, 64-bit, is required. Python is bundled and is not required on the user’s computer.

Trust Center

Trust Center does not transmit the active dataset. It provides a local sensitive-data scan that identifies likely email addresses, telephone numbers, IP addresses and payment-card patterns without retaining detected values. Findings are signals for review, not a legal classification or automated access-control decision.

Data contracts let an analyst define repeatable acceptance criteria. The initial rule set supports populated-value, uniqueness, numeric range, allowed-value, regular-expression and timestamp-freshness controls. Every rule has an explicit severity, and recommended rules are visible and editable before use. Running a check produces a weighted quality score together with Trusted, Needs attention, Blocked, or Not configured status. A data mutation invalidates the prior result automatically so stale evidence is never presented as current.

Output	Purpose
Trust Center table	Review each control, observed condition, expectation, violations and error state.
JSON audit evidence	Attach deterministic, portable validation evidence to an approval process, ticket or archive.
Styled HTML analysis report	Includes the current Trust Center summary and control results when checks were run.
.dsproj project	Preserves the contract definition; DataSense intentionally asks the user to rerun validation after restoring a project.

Included platform capabilities

DataSense remains a full desktop studio for data import, quality profiling, preparation, statistics, visualisation, SQL analysis, time series, machine learning, AutoML, dashboards, local project files and styled reports. The release pipeline builds the Windows x64 bundle, portable archive and installer from the tagged source revision; it also runs regression tests before publishing release assets.

Validation performed

The release candidate passed the complete automated suite, including dedicated contract, sensitive-data scanning, project persistence and off-screen desktop startup coverage. The PyInstaller recipe was additionally exercised in a Linux validation environment; Windows release artifacts are built only by the dedicated windows-latest GitHub Actions pipeline.

Upgrade note

Open existing projects normally. Their datasets and earlier analysis features are preserved. Existing projects do not contain a data contract until one is created in Trust Center.

Full Changelog: v2.1.0...v2.2.0

DataSense v2.1.0

DataSense 1.0.0

First commercial-grade release of DataSense — a desktop data analysis studio for Windows.

Downloads

File	Use it for
DataSense-1.0.0-setup.exe	Recommended installer (Start menu + desktop shortcut, uninstaller)
DataSense-1.0.0-windows-x64-portable.zip	Portable build — unzip and run DataSense.exe, no installation

What's inside

• Data workspace — import CSV, TSV, Excel, JSON, Parquet and SQLite; KPI overview, virtualised preview of large tables and a per-column quality profile (types, missing %, uniqueness, spread).
• Prepare workspace — fill or drop missing values, deduplicate, convert types, standardise/min-max scale, IQR outlier removal, computed columns, pandas-query filtering, group-by aggregation and pivot tables, all with full undo/redo history.
• Statistics workspace — descriptive statistics, Pearson/Spearman/Kendall correlation matrices, frequency tables, Shapiro-Wilk normality, two-sample and paired t-tests, one-way ANOVA, chi-square independence and OLS regression, each with a plain-language interpretation.
• Visualise workspace — ten chart types (line, area, bar, horizontal bar, scatter with trend line, histogram, box, violin, pie, correlation heatmap) with themed rendering, zoom/pan toolbar and PNG/SVG/PDF export.
• Machine learning workspace — regression and classification (linear/ridge/logistic, decision tree, random forest, gradient boosting, SVM) with train/test split, cross-validated scores, feature importance and confusion matrices, plus K-Means clustering and PCA; training runs off the UI thread.
• Reporting & projects — one-click styled HTML report of the session, .dsproj project files that store the dataset with its processing history, and dataset export to CSV/Excel/JSON/Parquet.
• Polished UI — custom dark and light design system, keyboard shortcuts, recent files, live status bar and a signed version-stamped executable.

Requirements

Windows 10 or 11 (64-bit). No Python installation required.

Full Changelog: v2.0.0...v2.1.0

DataSense v2.0.0

DataSense 1.0.0

First commercial-grade release of DataSense — a desktop data analysis studio for Windows.

Downloads

File	Use it for
DataSense-1.0.0-setup.exe	Recommended installer (Start menu + desktop shortcut, uninstaller)
DataSense-1.0.0-windows-x64-portable.zip	Portable build — unzip and run DataSense.exe, no installation

What's inside

• Data workspace — import CSV, TSV, Excel, JSON, Parquet and SQLite; KPI overview, virtualised preview of large tables and a per-column quality profile (types, missing %, uniqueness, spread).
• Prepare workspace — fill or drop missing values, deduplicate, convert types, standardise/min-max scale, IQR outlier removal, computed columns, pandas-query filtering, group-by aggregation and pivot tables, all with full undo/redo history.
• Statistics workspace — descriptive statistics, Pearson/Spearman/Kendall correlation matrices, frequency tables, Shapiro-Wilk normality, two-sample and paired t-tests, one-way ANOVA, chi-square independence and OLS regression, each with a plain-language interpretation.
• Visualise workspace — ten chart types (line, area, bar, horizontal bar, scatter with trend line, histogram, box, violin, pie, correlation heatmap) with themed rendering, zoom/pan toolbar and PNG/SVG/PDF export.
• Machine learning workspace — regression and classification (linear/ridge/logistic, decision tree, random forest, gradient boosting, SVM) with train/test split, cross-validated scores, feature importance and confusion matrices, plus K-Means clustering and PCA; training runs off the UI thread.
• Reporting & projects — one-click styled HTML report of the session, .dsproj project files that store the dataset with its processing history, and dataset export to CSV/Excel/JSON/Parquet.
• Polished UI — custom dark and light design system, keyboard shortcuts, recent files, live status bar and a signed version-stamped executable.

Requirements

Windows 10 or 11 (64-bit). No Python installation required.

Full Changelog: v1.1.0...v2.0.0

v1.1.0

Merge branch 'main' of https://github.com/Ali-Marandi/DataSense

DataSense v1.0.0

DataSense 1.0.0

First commercial-grade release of DataSense — a desktop data analysis studio for Windows.

Downloads

File	Use it for
DataSense-1.0.0-setup.exe	Recommended installer (Start menu + desktop shortcut, uninstaller)
DataSense-1.0.0-windows-x64-portable.zip	Portable build — unzip and run DataSense.exe, no installation

What's inside

• Data workspace — import CSV, TSV, Excel, JSON, Parquet and SQLite; KPI overview, virtualised preview of large tables and a per-column quality profile (types, missing %, uniqueness, spread).
• Prepare workspace — fill or drop missing values, deduplicate, convert types, standardise/min-max scale, IQR outlier removal, computed columns, pandas-query filtering, group-by aggregation and pivot tables, all with full undo/redo history.
• Statistics workspace — descriptive statistics, Pearson/Spearman/Kendall correlation matrices, frequency tables, Shapiro-Wilk normality, two-sample and paired t-tests, one-way ANOVA, chi-square independence and OLS regression, each with a plain-language interpretation.
• Visualise workspace — ten chart types (line, area, bar, horizontal bar, scatter with trend line, histogram, box, violin, pie, correlation heatmap) with themed rendering, zoom/pan toolbar and PNG/SVG/PDF export.
• Machine learning workspace — regression and classification (linear/ridge/logistic, decision tree, random forest, gradient boosting, SVM) with train/test split, cross-validated scores, feature importance and confusion matrices, plus K-Means clustering and PCA; training runs off the UI thread.
• Reporting & projects — one-click styled HTML report of the session, .dsproj project files that store the dataset with its processing history, and dataset export to CSV/Excel/JSON/Parquet.
• Polished UI — custom dark and light design system, keyboard shortcuts, recent files, live status bar and a signed version-stamped executable.

Requirements

Windows 10 or 11 (64-bit). No Python installation required.

Full Changelog: v1.0.1...v1.0.0

v1.0.1

Fix GitHub Actions permissions and update to latest versions