The following project shows how Exploratory Data Analysis (EDA) can be applied to an electricity usage data set. The project is created using Python within the Jupyter Notebook environment, including packages Pandas, Numpy, Matplotlib and Seaborn.
The National Energy Market (NEM) is the wholesale framework that generators and retailers use to trade electricity in Australia. We typically have the generators selling electricity to the retailers at a market price which is updated every hour. The retailers then hedge against the market, to distribute electricity to the consumers at a static price.
The marketplace uses an auction based system to allow generators to adjust prices when demand changes (peak hours), and/or when supply fluctuates (solar or renewables are limited).
Extremely large consumers can bypass the retailers and purchase electricity directly from the generators at the market price. It is these large consumers that have the opportunity to manage their electricity usage to provide monetary savings. For example
For this case study, we use the hypothetical example of the Alcoa Portland Aluminum Smelter in Victoria, Australia, and frame the data analysis as a business question;
How can Alcoa’s Portland Aluminum smelter use demand management to reduce its energy costs?
The aluminum smelter uses up to 10% of Victoria's total electricity demand, can purchase directly from generators and can typically use the following strategies to reduce electricity costs
The data set is taken from the Daily Electricity Price and Demand Data - Alex Kozlov (Kaggle)
https://www.kaggle.com/datasets/aramacus/electricity-demand-in-victoria-australia
It includes the total Daily Electricity Demand (MWh) in Victoria between 1st January 2015 and 6th October 2020 with the following features
Data preparation and cleaning is first applied to the dataset
First up we create a scatter plot to see how the price of electricity prices change with demand. There is a strong positive correlation, which is to be expected.
We can also create scatter plots that show how the electricity demand changes depending on the max/min daily temperatures. Interestingly we can see peaks at both high and low temperatures, and reduced demand at the milder temperatures at the center of the data charts.
After normalizing the data to account for there being more school days in our sample. We plot a histogram of the typical daily demand on school days versus non school days. We can derive the following from the chart
Similarly after normalizing our data to account for less public holidays we plot a histogram for usage on public vs non-public holidays;
We speculated that a higher solar index would correlate with lower prices due to excessive solar supply. However there is no visible correlation between solar index and price
https://www.kaggle.com/datasets/aramacus/electricity-demand-in-victoria-australia
Alex Kozlov - Daily Electricity Price and Demand datas
The code for this project can be found on my Github
Exploratory Data Analysis - Energy Pricing in Australia