ml-timeseries
Skill2 starsChanged 47 days agoMITView on GitHub
Setup de modelos de series temporais para saude. Skforecast, ARIMA, LSTM, Prophet. Triggers on /ml-timeseries.
In this repository26 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- abstractSkill
- artigoSkill
- author-creditSkill
- datasus-outcomeSkill
- desenho-metodologicoSkill
- graph-labSkill
- icSkill
- infograficoSkill
- journalSkill
- lab-expSkill
- latexSkill
- ml-checkpointsSkill
- ml-eval-reportSkill
- ml-pipelineSkill
- paper-pngSkill
- paper-reviewSkill
- paper-scaffoldSkill
- paperSkill
- papersSkill
- peer-reviewSkill
- radar-academicoSkill
- redSkill
- revisorSkill
- specSkill
- update-paperSkill
The file
Files
Skill: ml-timeseries
Pipeline especializado para previsao de series temporais em saude.
Bibliotecas preferidas
| Lib | Uso |
| skforecast | Forecasting com sklearn (LightGBM, XGBoost) |
| statsmodels | ARIMA, SARIMAX, decomposicao |
| prophet | Sazonalidade complexa, feriados |
| pytorch/LSTM | Sequencias longas, dados multivaridos |
| tslearn | Clustering de series |
Passos
1. Analise exploratoria
- Plot da serie completa
- Decomposicao: tendencia, sazonalidade, residuo
- Teste de estacionariedade (ADF, KPSS)
- ACF e PACF para identificar lags
2. Preprocessing
- Frequencia: garantir indice temporal regular (resample se necessario)
- Missing: interpolacao temporal (nao ffill/bfill cegamente)
- Outliers: IQR ou z-score, mas considerar surtos epidemiologicos como reais
- Diferenciacao se nao-estacionaria
3. Feature engineering temporal
- Lags: 1, 7, 14, 28 dias (ou equivalente da frequencia)
- Rolling: media e std movel (7, 14, 30 dias)
- Calendario: dia da semana, mes, feriado, estacao
- Exogenas: clima, campanhas de vacinacao, eventos de saude publica
4. Modelagem
Abordagem escalonada:
- Baseline: naive (ultimo valor), media movel
- Estatistico: ARIMA/SARIMAX
- ML: skforecast com LightGBM (ForecasterAutoreg)
- Deep: LSTM se dados > 1000 pontos
Validacao: TimeSeriesSplit ou expanding window. NUNCA shuffle em series temporais.
5. Metricas
- MAE, RMSE, MAPE
- Coverage do intervalo de confianca
- Plot: observado vs previsto com banda de confianca
6. Forecast
- Horizonte definido pelo usuario
- Intervalos de confianca (80% e 95%)
- Salvar previsoes em CSV com timestamps
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.