Model 2.1

Model 2.1

Clustered-meter average monthly ratios

Monthly (kWh) consumption clustering:

1. Twelve clustering models (one per month of signup): Spectral Clustering with nearest neighbors.

2. Clustering using (daily kWh) consumption extracted features. Cluster count set to 2-4: final count is decided by maximum silhouette score of manhattan distances.

3. Meters not in the training clustering data, are classified to a specific cluster using a Gaussian Process Classifier (GPC) on the same dataset (12 classifiers total).

Features (170) from weekend/weekday and full series:

- Statistical: median, variance, quantiles, ...

- Time-series: autocorrelations, trends, seasonalities, ...

Forecasting for a meter: Using the mean montly consumption + in-cluster month-to-month ratios.