BQML Multivariate Timeseries Forecasting Demo

Поділитися
Вставка
  • Опубліковано 26 жов 2024

КОМЕНТАРІ • 1

  • @AugustExner
    @AugustExner 8 місяців тому

    Not sure if you can helpt with this, but lets say i have some marketing data variables such as 'Cost' (Amount of money spent on marketing) and another variable 'Conversions_value' (how much have been sold). I train the model on data from month1 but in month2 i am interested in knowing how much my 'cost' should increase with in order to see a 'Conversions_value' increase with lets say 15%. My plan of attack would be the following:
    1. Create model and train on data from month1 with 'Cost' as my target variable.
    2. Create "new dataset" where i increase the "conversion_value" with 10%
    3. Use the ml.forecast function from the model with the "new dataset".
    Is this correctly understood?