ENERGIA GROUP

EV Customer Segmentation

K-Means & PAM clustering on EV consumption data, producing actionable customer segments for targeted marketing.

Python
scikit-learn
Clustering

Problem

Treating every electric vehicle customer the same makes targeted marketing difficult. The goal of this project was to move from raw EV consumption data to actionable customer segments that marketing activity could be tailored around.

TODO(Abraham): add the business context — who needed the segments, what marketing decisions they informed, and what (if anything) was in place before this work.

Approach

Segmentation was built in Python using two complementary clustering techniques on EV consumption data:

K-Means — partitions customers into k groups by minimising the distance between each customer and their cluster centroid.
PAM (Partitioning Around Medoids) — a related technique that uses actual customers (medoids) as cluster centres, making it more robust to outliers than centroid-based methods.

TODO(Abraham): describe the dataset — source, size, time period covered, and the features engineered from the consumption data.

TODO(Abraham): describe how the number of clusters was chosen and how the K-Means and PAM results were compared and validated.

Tech stack

Python
scikit-learn
K-Means and PAM clustering

TODO(Abraham): list any supporting libraries used (e.g. for data preparation or visualisation).

Outcome

The clustering produced actionable customer segments that could be used for targeted marketing.

TODO(Abraham): add the number of segments identified and a short description of what characterised each one.

TODO(Abraham): add any measurable outcome — e.g. how the segments were used, campaign results, or academic result if this was a study project.