House Hold Energy Data - Time Series

Introduction

The data was collected from my apartment unit in San Jose for one plus year. The data is collected with smart meters and shared by the energy company. This is time-series data by nature and can be used for various time-series Machine Learning experiments.

Description of Data

The data contains eight attributes.

TYPE - This is an information column. The value is 'Electric usage' for all the observations.
DATE - Date of electric consumption. There is no timestamp in this field.
START TIME - Start time of the consumption.
END TIME - End time of the consumption
USAGE - Consumption in kWh
UNITS - This column denotes measurement unit. It is kWh for all the observations.
COST - Cost of consumption in $.
NOTES - Mostly an empty column

Acknowledgements

I want to thank my energy provider PG&E for sharing the usage data.

Inspiration

I started looking into the data inspired by our research paper 'Energy Consumption Forecasting for Smart Meters' https://arxiv.org/abs/1512.05979. The work was done while I was working in India with a vibrant team of Data Scientists. During the time, we found that there was very little data available in the Public domain for the selected use case.