⚡ Free2move – Operations Analytics

Optimizing EV Charging Interventions

---

📌 Overview

This project explores how battery charging sessions and rental demand interact in electric vehicle operations.
The objective is to better understand when sending a field agent to manage charging operations still creates value, and when vehicles are likely to be rented before reaching full charge.

The analysis focuses on identifying battery level thresholds and time patterns where operational interventions may become unnecessary, helping reduce avoidable costs while maintaining vehicle availability.

📊 Dashboard Preview

Tableau : https://public.tableau.com/app/profile/alexandra.merli/viz/Classeur2_17665439841440/Tableaudebord1

🧠 Business Context

While vehicles are charging, there is a chance they may be rented before the battery reaches 100%.
In such cases, sending a field agent to move or manage the vehicle may not be cost-effective.

The key business question addressed is:

At what battery level does the probability of a vehicle being rented become high enough that further charging interventions no longer make sense?

---

🔍 Approach

The analysis follows a pragmatic, decision-oriented approach:

1. Data exploration

   • Charging sessions, battery levels at start and end
   • Rental events occurring during charging
   • Time-based patterns (hour of day)

2. Battery level segmentation

   • Battery levels grouped into 10% buckets to ensure interpretability
   • Focus on battery level at the end of charging sessions, as it is consistently available

3. Behavioral analysis

   • Rental likelihood during charging by battery level
   • Charging patterns by time of day
   • Agent service operations observed at different battery outcomes

4. Visualization & insights

   • Interactive Tableau dashboard to explore patterns and support operational decisions

---

⚠️ Key Assumptions & Limitations

• Some battery level start values related to agent operations are missing and could not be reliably reconstructed.
• Battery level at the end of the session is used as the most consistent indicator.
• The analysis prioritizes decision-making insights over perfect data completeness.

All assumptions are made explicit to avoid over-interpreting the data.

---

📦 Deliverables

• Interactive dashboard (Tableau)
• Exploratory analysis and transformations in Python
• SQL queries used for data preparation (BigQuery)
• Clear documentation of assumptions and methodology

---

🛠️ Tech Stack

• SQL (Google BigQuery)
• Python (data exploration & transformations)
• Tableau (visualization & dashboarding)

---

🎯 Outcome

This analysis highlights battery level ranges and time periods where charging interventions are less likely to be useful, providing a foundation for:

• optimizing agent dispatch decisions
• reducing unnecessary operational costs
• aligning charging strategies with actual rental demand

The project demonstrates an end-to-end analytics workflow, from raw data exploration to actionable operational insights.