Time series (Master of Science)

Two webinars (Week 1 and 3)

Plan

Week 1. Trend-seasonal decomposition and exponential smoothing models

* Data and Tasks
* Series Components
* Naive Models
	CHECK formula for srw forecast
* STL algorithm
* Series Characteristics
* ETS Model (Part I)
* ETS Model (Part II)
* Theta method

Week 2. ARIMA processes

* Stationary processes
* MA Process
* ARMA equation
* ARMA process
* ARIMA process
* SARIMA process
* Unit root tests: ADF test
* Unit root tests: KPSS test

Week 3. Forecasting

* Forecasting without a model
* More predictors!
* Predictors and ARIM A
* Model Quality
* Model Comparison
* Forecast comparison

Week 4. Pre-processing data

* Handling missing data
* Anomaly detection
* Structural break detection

Useful links:

https://robjhyndman.com/teaching/

https://ranalytics.github.io/tsa-with-r/

http://multithreaded.stitchfix.com/blog/2016/04/21/forget-arima/

https://cran.r-project.org/web/packages/Rlgt/index.html

https://cran.r-project.org/web/packages/dlm/index.html

Py:

https://juanitorduz.github.io/intro_sts_tfp/

https://github.com/ChadFulton/tsa-notebooks

https://towardsdatascience.com/structural-time-series-forecasting-with-tensorflow-probability-iron-ore-mine-production-897d2334c72b

http://num.pyro.ai/en/latest/tutorials/time_series_forecasting.html

Visualisation:

https://github.com/rstudio-conf-2020/dataviz

https://socviz.co/lookatdata.html

Causal Impact:

https://github.com/tcassou/causal_impact

...

https://www.quora.com/How-can-I-learn-Bayesian-time-series-analysis

https://github.com/uber/orbit

https://ocw.mit.edu/courses/economics/14-384-time-series-analysis-fall-2013/lecture-notes/

https://cran.r-project.org/web/packages/forecastML/vignettes/package_overview.html

https://business-science.github.io/modeltime/

https://github.com/awslabs/gluon-ts/

RNN прогнозирование получасового спроса на электричество в R: https://blogs.rstudio.com/ai/index.html#category:Time_Series

https://github.com/asael697/bayesforecast

https://github.com/asael697/varstan

https://github.com/business-science/modeltime.gluonts

данные с большим количеством нулей: intermittent demand with zeroes https://stats.stackexchange.com/questions/373689

тест Диболда-Мариано http://www.phdeconomics.sssup.it/documents/Lesson19.pdf

новый пакет для обработки дат https://cran.r-project.org/web/packages/clock/vignettes/clock.html

xenakas/om_ts_eng