Data Set Frequency

Data Set Frequency refers to the time frequency of the overall data set. The time frequency is set based on the target that has the most granular level data.

It is expected that the frequency of an entire data set remains constant across all targets. If a data set frequency is not constant across all targets, the most granular frequency target determines the overall data set frequency. The targets that are of a less granular frequency are based on the configured cleaning method selected in the Configure > Model page in the Model Build phase to get a complete series of the same frequency as the most granular data.

This is illustrated by the following two data sets.

The Left Dataset is the raw data given to Sensible Machine Learning. The Right Dataset is the data set processed by Sensible Machine Learning after the initial data load. In this scenario, Target A is a daily frequency and determines the underlying frequency of the entire data set. This means all other targets are expected to also be daily frequency. If any additional targets are not of the same frequency, Sensible Machine Learning fills their missing values based on the configured cleaning method selected in the Configure > Model page in the Model Build phase. Target B illustrates this as a weekly granularity in the Left data set and being filled with the mean value of the entire column in the Right Dataset.