Distribution of EV Charger Stations

Developing the most innovative solution for optimizing the distribution plan for EV charger stations in California. Utilized machine learning and applied business concepts, such as defining user personas, to create a creative and effective solution that was recognized by the top board managers of KPMG.

I would like to invite you to watch our YouTube video via this link: https://youtu.be/bwU8ZQm6Io0

Feature Engineering

. Converted raw data to per capita attributes

. Built upon normalization of variables

. Interpretation:

1. High Index: larger per capita value
2. Low Index: lower per capita value

Correlations Among Variables

Data Visualization

Feature selection

Based on the graph: A Married Caucasian with high salary and education level who commutes more by public transportation is more likely to be an EV Buyer.

Data Modeling & Evaluation

Output Conversion

Output Visualization

limitations

1. Feasibility Study is not part of the scope of this project (i.e., proximity and access cost to power grids), thus further research is needed

2. Datasets retrieved are from reliable sources, yet are estimated, hence may not be reflective of the true population and will affect generalization of the model

3. Factors affecting locations of EV charging stations are not limited to features used in our model

4. Assumptions made In pre-processing and feature engineering stages may impact the underlying distribution of the dataset thus ultimately affect our result.

5. The fitness of the model may be impaired by the quality of the data collected