Summary

Large scale availability of time varying data originating in financial, economical and social systems has led to a rise in the interest in this class of algorithms. Understanding how to model and describe time series information is a fundamental skill in any data scientists toolkit.

In this lecture, we will introduce you to the libraries that Python provides for working with time series data. We’ll quickly review the ARIMA class of models before diving in detail into the more advance ARCH class of forecasting models. Both real and synthetic datasets will be used to illustrate the different kinds of models and their underlying assumptions.

Program

Pandas for Time Series
- DataFrames and Series
- Time Series
- Subsetting
- DataFrame Manipulations
- Merge and Join
- Timeseries transformations
- Resampling/fill methods
statsmodels for Time Series
- Statsmodels structure and filosophy
- Pearson Correlation
- Correlations of 2 timeseries
- Autocorrelation
- Partial Autocorrelation Function
ARIMA Models
- Moving Average Models (MA)
- Autoregressive Models (AR)
- Fitting ARIMA models with statsmodels
- Fitting Seasonal ARIMA models
GARCH models
- Heteroscedasticy
- Heteroscedastical Models
- Fitting ARCH Model
- Generalizations