Can machine learning techniques predict customer dissatisfaction? A feasibility study for the automotive industry
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The automotive industry is in the strongest competition ever, as this sector gets disrupted by new arising competitors. Providing services to maximum customer satisfaction will be one of the most crucial competitive advantages in the future. Around 1 Terabyte of objective data is created every hour today. This volume will significantly grow in the future by the increasing numberof connected services within the automotive industry. However, customer satisfaction determination is solely based on subjective questionnaires today without taking the vast amount of objective sensor and service process data into account. This work presents an industrial application that fills this lack of research and thus provides a solution with a high practical impact to survive in the tough competition of the automotive industry. Therefore, the work addresses these fundamental business questions: 1) Candissatisfied customers be classified based on data that is produced during every service visit? 2) Can the dissatisfaction indicators be derived from service process data? A machine learning problem is set up that compared 5 classifiers and analyzed data from 19,008 real service visits from an automotive company. The 105 extracted features were drawn from the most significant available sources: warranty, diagnostic, dealer system and general vehicle data. The best result for customer dissatisfaction classification was 88.8% achieved with the SVM classifier (RBF kernel). Furthermore, the 46 most potential indicators for dissatisfaction were identified by the evolutionary feature selection. Our system was capable of classifying customer dissatisfaction solely based on the objective data that is generated by almost every service visit. As the amount of these data is continuously growing, we expect that the presented data-driven approach can achieve even better results in the future with a higher amount of data.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it