Course

Syllabus

List of Experiments

1. Exploratory Data Analysis (EDA) on Vehicle Telemetry Data – Analyze distribution, correlation, and outliers in real-time sensor data from a fleet of vehicles (e.g., temperature, pressure, RPM). Visualize trends and anomalies.
2. Predictive Maintenance Analysis – Predict the likelihood of part failure based on historical maintenance logs including repair dates, parts replaced, and mileage using classification algorithms like Decision Trees or Random Forests.
3. Fault Diagnosis using Sensor Data – Build a model to classify fault types based on Diagnostic trouble codes (DTCs) from onboard diagnostics (OBD) systems.
4. Anomaly Detection in Vehicle Performance – Detect anomalies in Performance metrics (e.g., fuel efficiency, engine power) across different driving conditions using statistical methods (e.g., Z- score, Isolation Forest) and visualize anomalies.
5. Image Recognition for Traffic Sign Detection – Develop a machine learning network model to detect and classify traffic signs (Images captured from vehicle-mounted cameras showing traffic signs) for autonomous driving applications.
6. Predictive Modeling for Vehicle Component Lifespan – Build a regression model to predict component lifespan of vehicle components (e.g., batteries, tires) based on usage and environmental conditions, using linear regression or survival analysis techniques.
7. Health Monitoring of Hybrid/Electric Vehicle Batteries – Develop a health monitoring system using time series analysis to predict battery degradation and failure using the Battery performance metrics (e.g., voltage, current, temperature) from hybrid/electric vehicles.
8. Driver Drowsiness Detection using Biometric Sensors – Build a classification model to detect driver drowsiness based on biometric sensor data (e.g., heart rate, eye movement collected from drivers during different driving conditions), using machine learning algorithms.

Course Objectives

• To introduce the concept of collecting and preprocess diverse automotive datasets including sensor data, maintenance logs.
• To impart hands-on experience in applying statistical analysis, regression, classification, clustering, and time series forecasting techniques to automotive data.
• To introduce complex automotive data problems, hypotheses, and data-driven solutions using a structured approach.

Course Outcomes

• CO01: Identify and address challenges specific to automotive data science.
• CO02: Acquire, clean, and preprocess diverse datasets from automotive sources, ensuring data readiness for analysis.
• CO03: Visualize data effectively using appropriate tools and interpret results to support decision-making in automotive diagnostics.
• CO04: Assess model accuracy, interpret results, and propose actionable recommendations based on predictive analytics and anomaly detection experiments.

CO-PO Mapping