This experiment is a demonstration on how user apply STL decomposition and ETS model on time series data in AzureML.
Data The data we are using here is from U.S. Energy Information Administration website. It is about weekly U.S. regular retail gasoline prices from 02-14-2011 to 02-09-2015. For more information, please check: http://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=pet&s=emm_epmr_pte_nus_dpg&f=w Experiment There are 3 parts in this experiment: training data, user input and training model. 1. Training Data: In this module, we input the training data to the experiment. 2. User Input: In time series prediction, we need to know how long in future user want to predict. This module allows user to put in number of days/weeks/months/years they want to predict in future. 3. Training Model: We first apply STL decomposition to the training data and then build ETS model. We take the input from user to define prediction horizon. The output of this module is the prediction values. Created by a Microsoft employee