EPIET/EUPHEM Time series analysis moduleArchived
The general content of the course provided the participants with basic theoretical concepts, using real life TS examples. The course included practical TSA exercises with STATA. Fellows learned how to describe and analyse a TS dataset. Fellows were also encouraged to discover if their current projects have a time dimension and if a TSA would be of added value for their project.
Objectives of the course:
- Understand the basic epidemiological applications of a time series (TS) (description, explanation, prediction) and how time series analysis can be used to interpret disease dynamics;
- Understand the theoretical principles that govern a time series (trend, periodicity, seasonality, white noise);
- Introduce the two main statistical techniques in TS analysis: autocorrelation and spectral analysis;
- Acquire basic practical skills related to time series analysis (handling the data, preparing the data for time series analysis, converting date formats, plotting the data, simple descriptive techniques, detecting and analysing outliers, dealing with missing values, performing stratified analysis, moving average and other smoothing techniques, regression models);
- Acquire basic computer skills (in STATA) and understand the respective commands in order to perform time series analysis (decompose the time series into its different components);
- Introduce the concept of stationarity and application of transformations, understand the use of filters, periodograms, correlograms, and acquire the basic computer skills needed to analyse these;
- Introduce the concept of using one time series to explain another;
- Understand how time series analysis can be used to forecast, predict, and set thresholds for outbreak detection, and acquire the basic computer skills needed to perform these operations.
The training course was held at the Spanish National School of Public Health in Madrid.