How playpowerlabs/edoptimize works
Project Type and Purpose
• This is an open-source R/Shiny-based learning analytics suite.
• It consists of three independent but structurally identical Shiny dashboard applications (Platform Analytics, Curriculum Analytics, Implementation Analytics), plus a shared data-prep component.
• It processes two raw data formats (event logs + assessment/item responses) into consumable RData artifacts, then serves them via interactive dashboards.
Top-Level Architecture Overview
• Data Ingestion Layer
– Raw Event Log & Raw Item Response Log (CSV or RData)
– data_prep/raw_data_simulation.R (optional simulator)
• Data Processing Layer (per dashboard)
– *_analytics/data/*_data_prep.R scripts
– Outputs: dashboard-specific RData files under each */data/ folder
• Application Layer (per dashboard)
– app.R (assembles UI + Server)
– appCSS.R (styling)
– functions.R (business logic + data wrangling helpers)
– plot.R (visualization helpers)
– ui/ … (Shiny UI modules, one file per tab)
– server/ … (Shiny server modules, one file per tab)
• Presentation Layer
– Shiny front-end delivered via Shiny Server or shinyapps.io
– End users interact via browser
Main Components & Their Relationships
A. Raw Data Sources
– Label: “Event Log”, “Item Response Log”
B. Data Prep Scripts (Orchestrator)
– Label: “platform_analytics_data_prep.R” (and analogous scripts)
C. Processed Data Store
– Label: “RData files”
– Loaded by app.R at startup
D. Dashboard Apps (three parallel apps)
– For each of Platform, Curriculum, Implementation:
• app.R (entry-point)
• UI Modules (ui/*.R)
• Server Modules (server/*.R)
• Shared functions.R & plot.R
E. User Interface & Interaction
– Label: “Web Browser (Shiny UI)”
Architectural Patterns & Design Principles
• Modularization: Each dashboard is composed of Shiny modules—one UI + one server file per tab.
• Separation of Concerns:
– Data Prep ↔ Business Logic (functions.R) ↔ Presentation (plot.R + UI)
• Reusability: Shared helper scripts (functions.R, plot.R) across dashboards.
• Stateless Frontend: All data is pre-computed; Shiny server only reacts to UI inputs.