This session will look at how we deal with time related data, and the different types of variation within it. Different methods will be discussed to identify and smooth out the affects of variation, with a view to being able to use data to predict future activity.
Defining time series data. Secular trend, cyclical variation, seasonal variation, residual variation. Linear and non-linear trends. Transformation of data. Rolling averages, weighting data, smoothing techniques including exponential smoothing. ARIMA techniques, BATS and TBATS. The relationship with process control and regression models