Overview

This project is for generating analysis, insights and discoveries in a DuckDB database.

Stack:

  • FastAPI
  • Pydantic to aid standardizing complex request / response objects
  • Python environment is '~/.pyenv/versions/nominates/bin/activate'

Goals:

Part 1 - Building the internal model (command-line driven)

  • Build an internal model of the data within the database; look for correlations, outliers, patterns etc ...
  • We'll use that model to assist us in a deep analysis across a subset (segment) of the data.
  • The output product is an AI-aided analysis dataset with insights, statistical information, correlations.
  • The internal model is serialized to disk when process completes.

Part 2 - Analayzing a subset of the data (FastAPI)

  • Users will then be able to send queries to the model for a particular database
  • For MVP, user sends array of unique Id's representing their segment.
  • A standard response object is generated (AI Directed element selection)
    • Statistical Analysis
    • Correlations
    • Outliers
    • Array of Sparkline summaries with associated data

Keywords:

Complex data slices, Multi-dimensional analysis, Cubed, AI-aided insight,

Data Source:

./data/pocket.db - Database source for modeling (symlinked)