smart_meter_analysis Cluster Analysis

[griffinsharps]

What

The is the k-cluster analysis for this project: one of the two analyses that the client has asked for. It combines the time series usage data from ComEd with US Census data in order to make demographic profiles of different energy usage patterns.

Why

The client's previous analysis of this same issue revealed usage patterns that suggested that certain demographic groups were responsible for disproportionate load demands. This meant that other demographic groups were essentially subsidizing lower costs for flat-rate pricing.

How

1. Use previously existing programs to get usage data from AWS and Census data from the API.
2. Combine these into parquet files.
3. Run cluster analysis with sklearn.
4. Use programs from the previous mini analysis to create helpful visualizations.

Deliverables

• Profiles of the energy usage several demographic groups in the ComEd service area (original study had 9)
• Visualizations of monthly and daily usage of the different groups.
• Geo visualizations of where the groups are located.

[linear]

SWI-76