Skip to content
Visory LLC

Data Science & Analytics Specialist

Full source: docs/ai-experts/06-data-science-analytics-expert.md

Scope

Section titled “Scope”

Statistical modeling, time-series forecasting (Prophet), and interactive visualization (Plotly primary, Altair, Matplotlib fallback) for analytics-first deliverables under projects/*/notebooks/ and projects/*/src/*/analytics/.

Key patterns

Section titled “Key patterns”
  • Prophet requires a Pandas DataFrame with ds/y columns — data is typically aggregated in Polars first, then converted (daily.to_pandas()).
  • Per-segment forecasting: loop over a category (e.g. sub_producto_id), skip segments with fewer than 30 observations.
  • Standardized Plotly styling helper: apply_standard_style() in projects/solar_wind_generation/src/solar_wind_generation/visory_viz_module.py.
  • Prefer Polars for datasets ≥10M rows; PyArrow-backed Pandas (convert_dtypes(dtype_backend="pyarrow")) for faster/lighter operations otherwise.

When to consult

Section titled “When to consult”
  • Implementing forecasting models
  • Creating interactive dashboards/charts
  • Building analytics pipelines in projects/*/analytics/
  • Optimizing data visualizations

See also: Projects for the specific analytics projects this persona’s examples are drawn from (solar_wind_generation, mx_com_reporting).