Data Preparation

In this stage the objective is preparing the data set for the Train and Execute phase. In particular, Data Preparation is the process of collecting, cleaning, transforming, and organizing data to make it usable. This process involves several steps, including the collection of raw data from various sources, cleaning it from inconsistencies, transforming it into formats suitable for analysis, and finally organizing it into a structured system.

Moreover, if there are some missing values and the seasonality is computed, the issue is managed by the model by replacing them with the average seasonality of the variable. Otherwise, if the seasonality has not been found, the missing values are automatically filled by the model with a linear interpolation.

The duplicate values are managed using an aggregation method that sums up all the values.

Each of the time series (both exogenous and endogenous) is standardized.

The same data preparation for the Train phase also applies for the Execute phase even if in the second one the training data are used again to have more feasible data preparation.