Model-agnostic interpretability of Deep Learning models for car emissions assessment
from SHAP explanations to behavioural archetypes and targeted recommendations
Joint work with A. Mancuso, R. Simone and F. Palumbo
2026-06-03
Drive green, drive safe
Sustainability
Aggressive drivingcan increase fuel consumption and CO₂ emissions by up to 40% (McConky, Chen, & Gavi, 2018; Xu, Li, Liu, Rodgers, & Guensler, 2017).
Green drivingimproves efficiency and reduces emissions (Zhou, Jin, & Wang, 2016).
Safety
Aggressive drivingis correlated with higher crash risk (Adavikottu & Velaga, 2021).
Structured green driving programshave achieved up to 10% fuel savings and a 33% reduction in property-damage accidents (Nævestad, 2022).
How to measure, explain, and improve driving behavior at scale?
Green-driving assessment
Ecoscoring project
A partnership between an Italian insurance company and University of Naples Federico II
Project objective
Develop an eco-scoring system to:
- assess driving behavior
- incentivize eco-friendly practices
- provide insurance premium discounts
Our focus in this work
Design an interpretable emission recommendation framework that:
- explains the drivers of emissions
- identifies behavioural archetypes
- generates targeted eco-driving recommendations
This system is complementary to the eco-score and enables actionable intervention.
Roadmap
from raw telematics to actionable recommendations
Data reconstruction and enrichment derive acceleration, slope, and environmental context from 2-minute GPS data
DL-based estimation of microscopic CO₂ emission rates
Customer-level surrogate modelling
structural + behavioural + contextual drivers
Model-agnostic explanations
individual drivers and global effect shapes
Behavioural archetypes
segmentation of explanation patterns → targeted recommendations
Data reconstruction and enrichment
Data sources
Boxes Data
Telematics devices
- 100,000 customers
- 6 months
- 600 million records
Recorded features
- timestamp
- latitude & longitude
- speed
- distance between recordings
- data collected every 2 minutes
customers data (provided by the insurance company)
- vehicle characteristics (fuel type, power, segment, registration year)
- customer characteristics (age class, geographical area)
main structural limitation
Two-minute resolution is too coarse to directly observe microscopic driving dynamics (acceleration, harsh braking, slope effects).
First idea: contextual driving style via routing APIs
The method
- start from black-box GPS + speed
- query routing services (
Open street maps/Google maps)
- reconstruct context (traffic, road type, expected travel time)
- compare expected vs actual → infer “aggressiveness”
Why it’s attractive
- context-aware (traffic + road class)
- behaviour measured relative to surroundings
- intuitive for insurers and policymakers
The bottleneck
At our scale, continuous API calls are too slow (Open Street map) or too expensive (Google Maps).
The API wall (reality check)
Free option
- ~600,000,000 records
- even at 1 request/second: \[600{,}000{,}000 \text{ seconds} \approx \mathbf{19 \text{ years}}\]
Need for plan B
External APIs are unscalable for moderately high frequency telematics.
DL-based estimation of microscopic CO₂ emission rates
Plan B: estimate emissions from vehicle physics
Key point
Emissions are driven by vehicle dynamics and resistive forces, not only distance.
Vehicle Specific Power (VSP)
\[VSP_t = \frac{A v_t + B v_t^2 + C v_t^3 + m v_t (a_t + g \sin\theta)}{m}\]
A: rolling resistance; B: rotation resistance; C: aerodynamic drag.
\(m\) is the vehicle mass, \(g\) is the gravity;
We needed: speed (\(v_t\)), acceleration (\(a_t\)), road grade (\(\theta\))
Plan B: physics-based emission models
a validated physical model
The gold standard is MOVES
(Motor Vehicle Emission Simulator, US EPA). (Koupal, Cumberworth, Michaels, & Beardsley, 2003; Park, Lee, & Lee, 2016)
MOVES pros
- Regulatory model
- Scientifically validated
- Physics-based
- Legally defensible
Why not MOVES?
MOVES cons
MOVES is designed for macroscopic studies (cities, regions)
Heavy database infrastructure
13+ complex input files
evaluating a single 10-minute trip: 2–5 minutes of processing
for hundreds of thousands of trips: computationally infeasible
NeuralMOVES: scalable surrogate of MOVES (Ramirez-Sanchez et al., 2026)
NeuralMOVES is a set of specialized neural networks trained on vehicle-specific subsets of the MOVES simulation data.
Why NeuralMOVES
Trained on millions of MOVES-generated scenarios
Tiny (MB-scale package)
Millisecond-level evaluation
highly accurate
NeuralMOVES architecture
Architecture
Inputs: speed, acceleration, grade
Context: temperature, humidity
2 hidden layers
5 neurons each
Activation:
hyperbolic tangent
rationale
Separate models by vehicle type
Avoid unrealistic mixing
Fast execution
Smooth, differentiable output
pre-processing and required features
Training
70/30 train/test
Early stopping (~300 epochs)
avoid negative emissions:
\[ \text{NeuralMOVES}(x) = \max\{\hat{e}(x), e_{idle}\} \]
NeuralMOVES required inputs
reconstruct from boxes data:
road slope
air temperature & humidity
acceleration
enrichment
Enrichment process
- slope: Computed from altitude changes along the GPS path.
- weather: External meteorological dataset (MeteoWeb.it); assigned by nearest province and month.
- acceleration: \(\frac{Speed_{t+1} - Speed_t}{\Delta t}\) between consecutive records.
Customer-level surrogate modelling
From emissions to explanations
what’s next?
customer-level understanding
key emission drivers
actionable guidance
Learning task
model NeuralMOVES-estimated emissions as a function of structural, behavioural, and contextual features.
Surrogate model
setup
observations: box-recording -> trip -> customer
target: log-transformed NeuralMOVES estimated emissions per km
- aggregation of trip-level \(\text{CO}_2\_{\text{g/km}}\) to obtain customer-level target values
vehicle & driver context features
- fuel / segment / power
- registration year (emission standards proxy)
- age class
- geography / area
driving behaviour features
- trip duration / distance structure
- speed level (median/mean)
- speed variability
- share of short trips
Candidate models
baseline
- Elastic Net (Zou & Hastie, 2005)
Tree-based models
- Random Forest (Breiman, 2001)
- XGBoost (Chen & Guestrin, 2016)
Surrogate model performance (RMSE, cross-validation)
Takeaway
XGBoost achieves the lowest RMSE (≈18% improvement over Random Forest), while Elastic Net underfits — nonlinear behavioural effects are essential.
Local model-agnostic explanations
intuition: from linear regression to SHAP (Lundberg & Lee, 2017)
Linear regression (observation \(i\))
\[ \hat y_i = \mathbb{E}[\hat y] + \sum_{j=1}^{p} \hat{\beta}_j \left( x_{ij} - \mathbb{E}[X_j] \right) \]
Each term \(\hat{\beta}_j\left(x_{ij}-\mathbb{E}[X_j]\right)\) is the contribution of feature \(j\) for observation \(i\), measured as a deviation from the average prediction.
SHAP decomposition (any model)
\[ f(\mathbf{x}_i) = \phi_0 + \sum_{j=1}^p \phi_{ij} \]
Each \(\phi_{ij}\) is the SHAP value associated with feature \(j\) for observation \(i\).
Why is it called SHAP?
Cooperative game theory
SHAP stands for:
SHapley Additive exPlanations
It is based on Shapley values from cooperative game theory (Shapley, 1953).
game theory
The Shapley value is a method for assigning payouts to players depending on their contribution to the total payout. Players cooperate in a coalition and receive a certain profit from this cooperation (Molnar, 2025).
mapping
- Game → prediction function \(f\)
- Players → features
- Coalition \(S\) → subset of features
- Payoff → model prediction
Shapley value
Formal definition
For observation \(i\) and feature \(j\):
\[ \phi_{ij} = \sum_{S \subseteq N \setminus \{j\}} \frac{|S|! (p-|S|-1)!}{p!} \Big[ f_{S\cup\{j\}}(\mathbf{x}_i) - f_S(\mathbf{x}_i) \Big], \]
where \(N = \{1,\dots,p\}\) and
\(f_S(\mathbf{x}_i)=\mathbb{E}[f(X)\mid X_S=x_{iS}]\).
Main intuition
The Shapley value is the average marginal contribution of feature \(j\) to the prediction for observation \(i\), averaged over all possible feature orders.
Weight
The term
\[ \frac{|S|!(p-|S|-1)!}{p!} \]
is the probability that subset \(S\) appears before feature \(j\) in a random ordering of features.
Marginal contribution
For a subset \(S\), the quantity
\[ f_{S\cup\{j\}}(\mathbf{x}_i) - f_S(\mathbf{x}_i) \]
measures how much feature \(j\) changes the prediction when added to the features in \(S\).
From Shapley theory to XGBoost
Structure of gradient boosting
An XGBoost model is additive:
\[ f(\mathbf{x}_i) = f_0 + \sum_{t=1}^{T} \lambda \, f_t(\mathbf{x}_i), \]
where:
- \(f_0\) = initial prediction (constant baseline)
- \(f_t\) = regression tree at iteration \(t\)
- \(\lambda\) = learning rate
Each tree is fitted to the pseudo-residuals.
SHAP and boosting
Shapley values are linear:
\[\begin{equation*} \begin{split} \phi_{ij}(f + g) &= \phi_{ij}(f) + \phi_{ij}(g)\\ \phi_{ij}(a f)&= a \ \phi_{ij}(f) \end{split} \end{equation*}\]
Therefore,
\[ \phi_{ij}(f) = \sum_{t=1}^{T} \lambda \, \phi_{ij}\!\left(f_t\right). \]
SHAP values can be computed tree-by-tree and then aggregated.
Why a single tree is special
Active path intuition
A regression tree partitions the feature space and assigns a constant prediction at each leaf.
For each observation \(\mathbf{x}_i\), only one root-to-leaf path is active.
Therefore, only the variables used along that path can affect the prediction for that tree.
TreeSHAP payoff
Instead of averaging over all possible feature subsets by brute force, TreeSHAP:
- follows the active path
- accounts for alternative paths induced by missing/present features
- aggregates marginal contributions efficiently
This makes exact SHAP values feasible for tree ensembles such as XGBoost.
Global SHAP importance
What drives predicted emissions?
Takeaway
Emissions are not explained only by vehicle structure:
they are also shaped by how people drive.
Global SHAP importance
What drives predicted emissions?
Behavioural archetypes
Towards actionable recommendations
From explanations to behavioural profiles
SHAP provides individual-level explanations.
For each customer \(i\) we obtain a SHAP vector
\[ \boldsymbol{\phi}_i = (\phi_{i1}, \dots, \phi_{ip}) \]
describing how behavioural variables contribute to predicted emissions.
The scalability problem
With 100,000 customers, SHAP yields 100,000 explanations.
Insurers cannot design 100,000 personalized interventions.
Instead, we need behavioural profiles that summarize recurring emission-driving patterns.
Idea
Cluster explanations, not customers.
We apply Archetypal Analysis to the SHAP vectors.
Archetypal Analysis on SHAP vectors
Data representation
Each customer is described by a SHAP explanation vector \(\boldsymbol{\phi}_i = (\phi_{i1},\dots,\phi_{ip})\)
Collecting them yields the SHAP matrix
\[ \boldsymbol{\Phi} = \begin{bmatrix} \boldsymbol{\phi}_1 \\ \vdots \\ \boldsymbol{\phi}_n \end{bmatrix} \in \mathbb{R}^{n\times p}. \]
Archetypal Analysis on SHAP vectors
Archetypal decomposition (Cutler & Breiman, 1994)
Archetypal Analysis approximates
\[ \boldsymbol{\Phi} \approx \mathbf{A}\mathbf{B}\boldsymbol{\Phi} \]
where
\(A_{n\times k}\) : customer → archetype mixture weights
\(B_{k\times}\) : archetype → customer mixture weights
Rows of both matrices are non-negative and sum to one.
Interpretation of archetypes
Archetypes on SHAP
Archetypes are the rows of
\[ \mathbf{Z} = \mathbf{B}\boldsymbol{\Phi} \]
Each archetype is therefore a convex combination of SHAP vectors:
\[ \mathbf{z}_h = \sum_{\ell=1}^{n} B_{h\ell}\boldsymbol{\phi}_\ell \]
Interpretation
Archetypes represent extreme emission mechanisms,
i.e. extreme patterns of feature contributions to predicted emissions.
Each customer is expressed as a convex combination of archetypes
\[ \boldsymbol{\phi}_i \approx \sum_{h=1}^{k} A_{ih}\mathbf{z}_h \]
→ enabling behavioural segmentation and targeted recommendations.
Why Archetypal Analysis?
Clustering vs archetypes
K-means
- by-group average behavioural profiles: overlap leads to similar profiles
Archetypal Analysis
- identifies extreme behavioural profiles
- observations are mixtures of archetypes
Selecting the number of archetypes
Archetype A1 — Structural vehicle disadvantage
Takeaway
Predicted emissions are mainly increased by vehicle age and fuel-type effects, while trip-structure variables partly offset the prediction. Speeding behaviour remains a secondary behavioural lever
Archetype A2 — Duration-intensive profile
Takeaway
Emissions are driven upward by sustained trip duration and vehicle age, partially offset by average trip length and speed-related effects
Archetype A3 — Short-trip inefficiency
Takeaway
Very short average trip length is the dominant positive contribution, consistent with a short-trip inefficiency regime
Archetype A4 — Long-duration compensated profile
Takeaway
Very long trip duration pushes emissions upward, but long average trip length and recent vehicle age compensate, producing a mixed profile
Archetype A1 — Structural vehicle disadvantage
Mechanism
Predicted emissions are mainly driven by structural vehicle characteristics
(e.g., vehicle age and technology).
Behavioural recommendations
Limited behavioural leverage; focus on vehicle efficiency improvements:
- Maintenance reminders (filters, tyres)
- Maintain correct tyre pressure
- Encourage efficient driving modes where available
Insurer action
Target vehicle upgrade incentives or eco-bonus programs rather than behavioural coaching
Archetype A2 — Duration-intensive profile
Mechanism
Long duration for moderatly long trips drives emissions upward
Behavioural recommendations
Prefer routes with less stop–go congestion
Encourage steady cruising instead of speed oscillations
Insurer action
Introduce long-trip coaching through periodic feedback and driving goals
Archetype A3 — Short-trip inefficiency
Mechanism
Frequent short trips create inefficiency due to cold starts and incomplete engine warm-up
Behavioural recommendations
- Trip chaining: combine errands into one trip
- Shift very short trips to active mobility
- Encourage smooth acceleration at trip start
Insurer action
Introduce a short-trip reduction challenge rewarding reductions in short-trip frequency
Archetype A4 — Long-duration compensated profile
Mechanism
Long trips tend to be efficient per km; the main risk comes from sustained high speeds
Behavioural recommendations
- Avoid very high sustained speeds
- Encourage anticipatory driving
- Maintain existing efficient driving habits
Insurer action
Provide lightweight feedback and eco-driving recognition incentives
From archetypes to scalable interventions
guidance
Instead of producing 100,000 individual recommendations,
we design archetype-specific guidance
This enables:
- nudging strategies
- targeted coaching
- eco-driving incentives aligned with emission mechanisms
Wrap-up
What we did
- reconstructed microscopic inputs from coarse telematics
- used NeuralMOVES to estimate high-frequency CO₂ emission rates at scale
- built an accurate customer-level surrogate (XGBoost)
- explained predictions with SHAP
- segmented explanation patterns via Archetypal Analysis to enable targeted recommendations
Wrap-up
Take-home messages
- interpretability is not only about explaining models — it enables segmentation and action
- applying AA to SHAP uncovers behavioural profiles aligned with emission mechanisms
- recommendations become scalable without generating 100,000 individual plans
Outlook
- integrate the recommendation layer into the eco-score pipeline
- assess robustness across vehicle segments, seasons, and geography
References
