Each dataset has a frequency. They can be daily, weekly, monthly, quarterly, or yearly, This step is used to change that frequency and recalculate the values of the variables.
With Change Frequency users can resample the time-frequency of the dataset moving from Daily to monthly, quarterly, and yearly back and forth. When doing so the user has to select the aggregation formula if moving from higher to a lower frequency - such as average or end of the period - or the interpolation formula if moving from lower to higher, such as linear interpolation or splines
Step 1. Create a pipeline and choose the data source
Step 2. add a Change Sample Frequency step. Once the step is selected the following Interface will appear. The system will automatically identify the input data frequency.
Step 3. Select the output data. This will define whether the step is an interpolation or Downscaling.
When going from a lower frequency to a higher frequency (i.e. monthly to daily) you have to select from the 8 interpolation options: Repeat the value, linear interpolation, quadratic, cubic, polynomial, and piecewise, Splines and Krogh interpolation.
Alternatively, when going from higher to lower (I,e. monthly to quarterly) there are 6 options: average all the values, sum, select the min or max, or the last and first.
These are examples of the different interpolation methods applied on the same input dataset.
https://charts.alphacast.io/p/Alphacast_Support/public-repo/alphacast-interpolation-options?tab=chart&stackMode=absolute&country=ARG-0~ARG-7~ARG-6~ARG-3~ARG-2®ion=World (opens in a new tab)
https://charts.alphacast.io/p/Alphacast_Support/public-repo/alphacast-interpolation-options?tab=chart&stackMode=absolute&country=ARG-0~ARG-8~ARG-4~ARG-1®ion=World (opens in a new tab)