Visualization
Interactive Dashboards & Data Storytelling
Visualizations that transform complex data into actionable insights.
Production Dashboards
Marketing Mix Modeling Dashboard
Framework: Shiny (R)
Interactive dashboard to visualize Marketing Mix Modeling results and simulate budget allocation scenarios.
Features:
- Sales decomposition by channel
- Response and saturation curves
- Budget simulator
- ROI comparison by channel
View Dashboard
Marketing Analytics Dashboard
Framework: Streamlit (Python)
Dashboard for tracking multi-channel marketing KPIs with real-time data from BigQuery.
Features:
- Global KPIs (Spend, Revenue, ROAS)
- Temporal trends
- Performance by channel
- Conversion funnel
View Dashboard
Clickstream Analytics (Local)
Framework: Streamlit + Plotly
Real-time dashboard for e-commerce behavioral analysis with streaming data.
Features:
- Real-time metrics
- Interactive funnel
- Top products
- Configurable auto-refresh
Available locally with Docker
See project →
Visualization Technologies
Streamlit
Python framework to quickly create interactive data applications.
Used in:
- Causal Inference
- Marketing Pipeline
- Clickstream Analytics
Shiny
R (and Python) framework for advanced statistical dashboards.
Used in:
- MMM Robyn
- Penguin Explorer
Plotly
Interactive charting library for Python and R.
Chart types:
- Scatter plots
- Time series
- Funnel charts
- Heatmaps
Quarto
Scientific publishing system for creating reports and websites.
Used for:
- This portfolio site
- Project documentation
- Analysis reports
Visualization Types
Marketing Analysis
-
Sales decomposition: Waterfall charts showing each channel’s contribution
-
Response curves: Diminishing returns visualization
-
Budget heatmaps: Optimal allocation matrix
Time Series
-
Trends: KPI evolution over time
-
Seasonality: Weekly/annual patterns
-
Anomalies: Visual outlier detection
Machine Learning
-
SHAP plots: Variable importance and local explanations
-
Posterior distributions: Bayesian uncertainty visualization
-
Conversion funnel: Sankey diagrams for user journeys
Bayesian Statistics
-
Trace plots: MCMC convergence
-
Forest plots: Credible intervals
-
Posterior predictive checks: Model validation
Design Principles
My visualizations follow these principles:
-
Clarity: One chart = one message
-
Context: Always show relevant comparisons
-
Interactivity: Enable exploration without overload
-
Accessibility: Color palettes adapted for all
-
Reproducibility: Code available for each visualization
← Back to Portfolio | See Analysis projects →