Baseline 5

Let's take an example, if we want to predict the missing months of the smart meters which signe up in May (we want to predict January, february, March and April), the following steps are applied : 1. We want to make a prediction for the smart meters that signed up in May (There are +/- 270 that sign up in May out of the 3258 smart meters) 2. Select all the smart meters that signed up in May or earlier (+/- 1350 smart meters = 270 x 5 smart meters = the 270 that sign up in January + the 270 in Feb + 270 in March + 270 in April + 270 in May) 3. Apply Kmeans clustering using DTW on the 1350 selected smart meters based on the data from May to December.4. To predict April : compute the average consumption for the month of April for each cluster, based on the smart meters which have data in April. 5. This average consumption is than repeated for the member of the cluster which have no data for April.