An overview of methods for analysis of numerical and categorical time series.
We will have meetup about time series analysis. It will be more educational than deep. The first part will be presented by Igor Itkin, a Data Scientist from http://www.sogoservices.com/. If you are interested in basic math and some of the most common algorithms used for analysis of numerical time series – this part is just for you.
The main topics are:
- Overview. What is so special about time series?
- Basic definitions. Trend, seasonality, slow and fast variation,Madrid, regularity, autocorrelation.
- Stationarity, scedasticity and some statistics.
- Predictions using time series. Cross validation.
- Averaging. Simple methods.
- More advanced methods (ARIMA and others).
The second part will be presented by David Gruzman (www.nestlogic.com).We will overview methods to analyze time series of categorical values. We will look at markov chains and various RNN (recurrent neural networks) (e.g. LSTM, GRU).Our goal will be to understand their applicability and potential value when processing time series.