Features

Complete product capability inventory

VizAdvisor ships as a full-stack advisor with optional statistical analysis. The inventory below mirrors the project README and CHANGELOG.

1 Product features

Feature	Description
CSV/JSON Upload	Drag-and-drop or paste data; automatic schema inference
Goal Selection	Compare, trend, distribution, correlation, part-of-whole, geospatial, network, ranking
LLM Recommendations	Structured output: chart type, rationale, design decisions, code scaffold
Pre-Viz Analysis	Optional descriptive stats, regression, power, mediation, factorial (R or Python)
Post-Viz Analysis	Same analysis types after receiving recommendations
Dark Mode	Theme toggle with persistent preference
Export	Copy or download recommendations as Markdown/JSON
Session History	Save and load past sessions (`localStorage`)

2 Output components

Component	Role
`RecommendationCard`	Primary recommendation, rationale, data mapping
`AlternativeOptions`	Alternative chart options
`DesignDecisionsPanel`	Color, scale, annotations, accessibility
`PitfallWarnings`	Pitfalls and mitigations
`FollowUpQuestions`	Suggested follow-up questions
`CodeSnippet`	Syntax-highlighted code with copy
`MetaBadges`	Confidence and goal category
`ExportButton`	Markdown / JSON export

3 Input components

Component	Role
`DataUploader`	File drop or paste
`DataPreview`	Schema table and type override
`GoalSelector`	Goal category + description
`ParameterPanel`	Audience, library, interactivity, accessibility, notes
`PromptBuilder`	Submit / reset and readiness gating

4 Analysis types

Type	Description
Descriptive	Summary stats, optional group-by
Regression	Linear regression / ANOVA
Power	Power analysis (u, v, f², power)
Mediation	X → M → Y mediation
Factorial	Factorial ANOVA (2–3 factors)

5 Supported chart libraries (scaffolds)

Recommendations can emit code for Recharts, D3.js, Plotly, Chart.js, Vega-Lite, Observable Plot, Matplotlib, Altair, and ggplot2 — selected via the parameter panel.

6 Theoretical foundations

Recommendations are grounded in:

Edward Tufte — data-ink ratio, chartjunk, small multiples
Tamara Munzner — What–Why–How framework, channel effectiveness
Alberto Cairo — truthful, functional, beautiful visualization
Stephen Few — perceptual efficiency and dashboard design
Colin Ware — visual perception and pre-attentive attributes
Claus Wilke — principles of figure design

See Data Viz Reference and Prompt Design for the full knowledge base encoded in the system prompt.

--- title: "Features" subtitle: "Complete product capability inventory" --- VizAdvisor ships as a full-stack advisor with optional statistical analysis. The inventory below mirrors the project README and CHANGELOG. ## Product features | Feature | Description | |---------|-------------| | **CSV/JSON Upload** | Drag-and-drop or paste data; automatic schema inference | | **Goal Selection** | Compare, trend, distribution, correlation, part-of-whole, geospatial, network, ranking | | **LLM Recommendations** | Structured output: chart type, rationale, design decisions, code scaffold | | **Pre-Viz Analysis** | Optional descriptive stats, regression, power, mediation, factorial (R or Python) | | **Post-Viz Analysis** | Same analysis types after receiving recommendations | | **Dark Mode** | Theme toggle with persistent preference | | **Export** | Copy or download recommendations as Markdown/JSON | | **Session History** | Save and load past sessions (`localStorage`) | ## Output components | Component | Role | |-----------|------| | `RecommendationCard` | Primary recommendation, rationale, data mapping | | `AlternativeOptions` | Alternative chart options | | `DesignDecisionsPanel` | Color, scale, annotations, accessibility | | `PitfallWarnings` | Pitfalls and mitigations | | `FollowUpQuestions` | Suggested follow-up questions | | `CodeSnippet` | Syntax-highlighted code with copy | | `MetaBadges` | Confidence and goal category | | `ExportButton` | Markdown / JSON export | ## Input components | Component | Role | |-----------|------| | `DataUploader` | File drop or paste | | `DataPreview` | Schema table and type override | | `GoalSelector` | Goal category + description | | `ParameterPanel` | Audience, library, interactivity, accessibility, notes | | `PromptBuilder` | Submit / reset and readiness gating | ## Analysis types | Type | Description | |------|-------------| | **Descriptive** | Summary stats, optional group-by | | **Regression** | Linear regression / ANOVA | | **Power** | Power analysis (u, v, f², power) | | **Mediation** | X → M → Y mediation | | **Factorial** | Factorial ANOVA (2–3 factors) | ## Supported chart libraries (scaffolds) Recommendations can emit code for Recharts, D3.js, Plotly, Chart.js, Vega-Lite, Observable Plot, Matplotlib, Altair, and ggplot2 — selected via the parameter panel. ## Theoretical foundations Recommendations are grounded in: - Edward Tufte — data-ink ratio, chartjunk, small multiples - Tamara Munzner — What–Why–How framework, channel effectiveness - Alberto Cairo — truthful, functional, beautiful visualization - Stephen Few — perceptual efficiency and dashboard design - Colin Ware — visual perception and pre-attentive attributes - Claus Wilke — principles of figure design See [Data Viz Reference](DATA-VIZ-REFERENCE.md) and [Prompt Design](PROMPT-DESIGN.md) for the full knowledge base encoded in the system prompt.