This notebook does a detailed analysis about ASHRAE's energy consumption data available in Kaggle.
Three csv files analyzed, in total 20 million rows:
- building_metadata: characteristics of a building
- weather_train: weather data measured between 1 Jan - 31 Dec 3016 among different sites
- train: energy consumption of the buildings available in building_metadata between 1 Jan - 31 Dec 3016 in four different energy categories:
- electricity
- steam
- chilled water
- hot water
This is also initial part of a machine learning problem aiming to predict the energy consumption of the building. Thus, during the detailed analysis, focus is kept on the meter_reading
values and its relationship to other variables.
Original kernel is available here.
Analysis is done with Python 3 and with the following libraries:
- plotly
- pandas
- seaborn
- pandas_profiling