Modelling Non-Equidistant Time Series Using Spline Interpolation*

E-mail: [email protected] This study gives an overview of spline interpolation, a special class of interpolation methods. The focal concern discussed in this paper is that the augmentation of non-equidistant time series (using averages, previous values, or interpolation) often leads to misleading or erroneous conclusions, as the augmented time series may have different characteristics than the original data generating process. The author’s main purpose is to demonstrate that augmentations of any kind are to be planned carefully. To underline this statement, he applies the most frequently used methods on empirical time series, then collects and highlights the most prevalent conclusions.