How nahianibnat-2000/esports-intelligence works
Esports Intelligence ingests Oracle’s Elixir professional League of Legends data, builds Elo and machine-learning comparisons, computes statistical analyses, and serves the results through a web app and API. The primary user flow is browser views or prediction requests → FastAPI domain services → stored match/model/analysis data; defaults can instead render from precomputed snapshots. Offline, the documented pipeline acquires, cleans, validates, transforms, and loads data before training and analytics. The graph preserves the major documented product areas; some subsystem links rely on README descriptions because their implementation was not sampled.
Main entry points: train.py Model training · core.py Platform database · run.py Ingest workflow · walk_forward.py Walk-forward studies
Components 24
Web experience
- Web views
frontend/app - API client
frontend/lib/api.ts - Page snapshots
frontend/lib/snapshot.ts
API and domain
- REST routes
backend/app/api/routes/v1.py - Domain services
backend/app/services - Research reports
backend/app/services/research.py - Provider adapters
backend/app/providers - DB sessions
backend/app/database/session.py - Platform database
backend/app/models/core.py
Data ingestion
- Oracle Elixir files
- Source acquisition
pipelines/oracle_elixir/download.py - Ingest workflow
pipelines/oracle_elixir/run.py - Clean and validate
pipelines/oracle_elixir/clean.py - Table transforms
pipelines/oracle_elixir/transform.py - Database loading
pipelines/oracle_elixir/load.py
Ratings and research
- Temporal splits
machine_learning/data_preparation/splits.py - Elo ratings
machine_learning/elo/elo.py - Prematch features
machine_learning/feature_engineering/features.py - Model training
machine_learning/training/train.py - Model evaluation
machine_learning/evaluation/metrics.py - Walk-forward studies
machine_learning/experiments/walk_forward.py
Statistical analysis
- Analytics batch
analytics/run_all.py - Balance measures
analytics/economics/concentration.py
Other components
- Platform user
Connections 28
- Platform user → Web views: browses
- Web views → API client: requests data
- Web views → Page snapshots: loads defaults
- API client → REST routes: calls API
- REST routes → Domain services: dispatches requests
backend/app/api/routes/v1.py - REST routes → Research reports: serves reports
backend/app/api/routes/v1.py - Domain services → DB sessions: uses sessions
- DB sessions → Platform database: connects
backend/app/database/session.py - REST routes → Provider adapters: uses adapters
backend/app/api/routes/v1.py - Oracle Elixir files → Source acquisition: provides files
- Ingest workflow → Source acquisition: acquires data
pipelines/oracle_elixir/run.py - Ingest workflow → Clean and validate: runs cleaning
pipelines/oracle_elixir/run.py - Ingest workflow → Table transforms: runs transforms
pipelines/oracle_elixir/run.py - Ingest workflow → Database loading: loads tables
pipelines/oracle_elixir/run.py - Database loading → Platform database: writes records
pipelines/oracle_elixir/load.py - Platform database → Model training: supplies match data
- Temporal splits → Model training: assigns periods
machine_learning/training/train.py - Model training → Elo ratings: replays ratings
machine_learning/training/train.py - Model training → Prematch features: builds features
machine_learning/training/train.py - Model training → Model evaluation: scores models
machine_learning/training/train.py - Walk-forward studies → Temporal splits: uses splits
machine_learning/experiments/walk_forward.py - Walk-forward studies → Elo ratings: compares ratings
machine_learning/experiments/walk_forward.py - Walk-forward studies → Prematch features: builds features
machine_learning/experiments/walk_forward.py - Walk-forward studies → Model evaluation: evaluates results
machine_learning/experiments/walk_forward.py - Model training → Platform database: persists models
- Analytics batch → Platform database: reads and stores outputs
analytics/run_all.py - Analytics batch → Balance measures: computes measures
analytics/run_all.py - Research reports → Analytics batch: uses computed outputs
backend/app/services/research.py