Skip to main content
CARLA HQ
Automotive & Fleet Management Automotive & Mobility use casesNACE G45.1Multiclass ClassificationTarget: car_acceptabilityOpenML #40975 ↗

Car

The Car Evaluation dataset comprises 1,728 vehicle profiles evaluated across 6 categorical attributes covering price, maintenance, and technical capacity. It serves as a canonical multi-class classification benchmark to predict overall vehicle acceptability across four discrete tiers.

✨ Try in CarlaInstall Chrome Extension ↗100% In-Browser WebGPU • Zero Cloud Upload
1,728Records (Rows)
6Predictive Features
0 / 7Numeric / Categorical
0.0%Missing Value Ratio
0.0231Log-Loss
1.48sWebGPU Mean Split
BUSINESS CONTEXT & OBJECTIVEAutomotive Retail & Fleet Management (G45.1)

Business Objective: Car

01Business Context

Automotive dealerships, fleet procurement managers, and appraisal platforms evaluate multi-attribute vehicle trade-ins and inventory specifications to determine market readiness, customer appeal, and valuation tiers.

02Analytical Objective

Accurately classify vehicle acceptability into four discrete market categories (unacceptable, acceptable, good, very good) based on physical attributes, operating costs, and safety ratings.

03Economic & Decision Impact

Optimizes inventory acquisition margins and turnaround times. Misclassifying an unacceptable vehicle as acceptable (false positive) introduces unsellable, high-depreciation inventory, while under-rating premium vehicles (false negative) forfeits competitive deal volume and customer satisfaction.

ML Benchmark Narrative

Due to its deterministic hierarchical rule origins from the DEX expert system, tree-based algorithms (Random Forest, LightGBM, Decision Trees) achieve near 98–100% accuracy once ordinal encodings are recognized. Tabular foundation models like TabICL on Carla HQ replicate these intricate non-linear decision boundaries zero-shot directly in spreadsheet environments without requiring explicit feature engineering or pipeline tuning.

Source Origin:UCI Machine Learning Repository & OpenML
Creator:Marko Bohanec and Blaž Zupan (1997)
License:Creative Commons Attribution 4.0 International (CC BY 4.0)
FEATURE SPECIFICATION & DATA DICTIONARY

Target & Feature Column Definitions

Exhaustive business definitions, measurement units, roles, and target variables across all 7 columns.

ColVariable / Feature NameRoleTypeUnit / ScaleDescription & Business Meaning
Acar_acceptabilityAcceptability RatingTARGETCAT
Target categorical acceptability rating: unacc (unacceptable), acc (acceptable), good (good), vgood (very good).
Bbuying_price_tierBuying Price TierFeatureCAT
Relative purchase cost level of the vehicle (vhigh: very high, high: high, med: medium, low: low).
Cmaintenance_cost_tierMaintenance Cost TierFeatureCAT
Relative ongoing maintenance and service expense level (vhigh: very high, high: high, med: medium, low: low).
Ddoor_countNumber of DoorsFeatureCATdoors
Categorical door count configuration of the vehicle (2, 3, 4, 5more).
Epassenger_capacityPassenger CapacityFeatureCATpassengers
Maximum passenger seating capacity category (2, 4, more).
Fluggage_boot_sizeLuggage Boot SizeFeatureCAT
Trunk and cargo storage capacity classification (small, med, big).
Gsafety_ratingEstimated Safety LevelFeatureCAT
Evaluated passenger safety level category (low, med, high).
DATASET PREVIEW

Interactive Data Table

Explore rows, feature values, and target labels for Car.

Loading dataset...
LOCAL EXECUTION TELEMETRY

TabICLv2 WebGPU Benchmark Results

Complete performance metrics on persisted IID or non-IID evaluation splits, executed 100% locally in the browser sandbox.

Mean Split Latency?Mean wall-clock execution time per persisted evaluation split running 100% locally via WebGPU inside Chrome.Explore Latency Guide →
1.48s30 splits • 8 ensembles/split
Total Evaluation Runtime?Total wall-clock duration across all selected splits, including in-context encoding, WebGPU shader execution, and result aggregation.Explore Split Protocol →
44.26sIncludes warmup & sync
Evaluation Protocol?Precomputed row indices ensure the browser and Python runners evaluate identical out-of-sample observations without leakage.Explore Split Protocol →
10×3 SplitsRepeated IID • 30 total
TabICLv2 Score?Primary task performance score (Log-Loss) achieved by TabICLv2 across out-of-sample evaluation splits.Explore Metric Formula →
0.0231Log-Loss
Evaluation SplitTrain RowsTest Rows
Split Duration?Wall-clock inference time taken for this evaluation split running 100% locally via WebGPU.Learn more →
Accuracy?Proportion of correct test predictions across all classes.Google ML Guide & Details →
ROC-AUC?Area under the ROC curve evaluating ranking capability across all thresholds.Google ML Guide & Details →
F1 Score?Harmonic mean of Precision and Recall, robust against class imbalance.Learn more →
Precision?Proportion of predicted positives that were actually positive.Google ML Guide & Details →
Recall?Proportion of actual positive cases successfully captured.Google ML Guide & Details →
R1 / F11,1525761.47s99.48%0.99910.93620.91670.9565
R1 / F21,1525761.46s99.13%0.99870.90200.82141.0000
R1 / F31,1525761.46s99.48%0.99870.93620.91670.9565
R2 / F11,1525761.45s98.61%0.99690.84000.77780.9130
R2 / F21,1525761.47s98.78%0.99800.85110.83330.8696
R2 / F31,1525761.46s98.96%0.99780.87500.84000.9130
R3 / F11,1525761.47s99.48%0.99970.93330.95450.9130
R3 / F21,1525761.47s98.09%0.99400.78430.71430.8696
R3 / F31,1525761.47s99.48%1.00000.93880.88461.0000
R4 / F11,1525761.47s99.48%0.99860.93880.88461.0000
R4 / F21,1525761.46s98.96%0.99750.88000.81480.9565
R4 / F31,1525761.46s97.74%0.99350.71110.72730.6957
R5 / F11,1525761.48s99.48%0.99870.93880.88461.0000
R5 / F21,1525761.52s98.26%0.99610.80000.74070.8696
R5 / F31,1525761.48s99.83%1.00000.97781.00000.9565
R6 / F11,1525761.46s99.31%0.99910.91670.88000.9565
R6 / F21,1525761.54s98.96%0.99740.88000.81480.9565
R6 / F31,1525761.48s98.61%0.99730.80000.94120.6957
R7 / F11,1525761.47s98.78%0.99540.86790.76671.0000
R7 / F21,1525761.47s99.48%0.99920.93330.95450.9130
R7 / F31,1525761.48s99.31%0.99910.91300.91300.9130
R8 / F11,1525761.51s98.96%0.99870.88460.79311.0000
R8 / F21,1525761.47s98.61%0.99720.83330.80000.8696
R8 / F31,1525761.48s98.78%0.99870.83720.90000.7826
R9 / F11,1525761.47s99.48%0.99920.93880.88461.0000
R9 / F21,1525761.47s98.78%0.99760.84440.86360.8261
R9 / F31,1525761.48s99.31%0.99860.91300.91300.9130
R10 / F11,1525761.45s99.31%0.99920.91300.91300.9130
R10 / F21,1525761.48s98.96%0.99810.87500.84000.9130
R10 / F31,1525761.48s98.61%0.99760.81820.85710.7826
Mean ± Std1.48s99.02% ± 0.47%0.998 ± 0.0020.880 ± 0.0590.8580.910
UNIFIED BENCHMARK LEADERBOARD

Foundation Models vs Traditional ML Baselines

Side-by-side evaluation of our in-browser Carla engine (WebGPU), open-source tabular foundation models, and OpenML baselines.

RankAlgorithm / ModelModel FamilyRuntime
Predictive Accuracy?Proportion of correct test predictions across cross-validation splits.Google ML Guide & Details →
ROC-AUC?Area under the ROC curve evaluating ranking capability.Learn more →
F-Measure?Harmonic mean of precision and recall.Learn more →
Reference / Repo
#1pipeline.Pipeline(imputation=openmlstudy14.preprocessing.ConditionalImputer,hotencoding=sklearn.preprocessing.data.OneHotEncoder,scaling=sklearn.preprocessing.data.StandardScaler,variencethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,classifier=sklearn.svm.classes.SVC)Support Vector Machine (SVM)Python100.00%1.00001.0000OpenML #9202815 ↗
#2MultiSearch(weka.classifiers.meta.multisearch.RandomSearch,weka.classifiers.meta.FilteredClassifier(weka.filters.MultiFilter(weka.filters.unsupervised.attribute.ReplaceMissingValues,weka.filters.unsupervised.attribute.RemoveUseless,weka.filters.unsupervised.attribute.Normalize),weka.classifiers.functions.SMO(weka.classifiers.functions.supportVector.RBFKernel,weka.classifiers.functions.Logistic)))Support Vector Machine (SVM)Python100.00%0.99921.0000OpenML #10416489 ↗
#3pipeline.Pipeline(imputation=hyperimp.utils.preprocessing.ConditionalImputer,hotencoding=sklearn.preprocessing.data.OneHotEncoder,scaling=sklearn.preprocessing.data.StandardScaler,variencethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,clf=sklearn.svm.classes.SVC)Support Vector Machine (SVM)Python100.00%1.00001.0000OpenML #9011260 ↗
#4Google TabFM v1.0Tabular Foundation ModelPyTorch/Python99.65%0.99950.9569google-research/tabfm ↗
#5pipeline.Pipeline(columntransformer=sklearn.compose. column transformer.ColumnTransformer(numeric=sklearn.pipeline.Pipeline(imputer=sklearn.preprocessing.imputation.Imputer,standardscaler=sklearn.preprocessing.data.StandardScaler),nominal=sklearn.pipeline.Pipeline(simpleimputer=sklearn.impute. base.SimpleImputer,onehotencoder=sklearn.preprocessing. encoders.OneHotEncoder)),variancethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,mlpclassifier=sklearn.neural network.multilayer perceptron.MLPClassifier)Neural Network (MLP)Python99.02%0.99980.9902OpenML #10418653 ↗
#6Carla Engine (TabICLv2 WebGPU)Tabular Foundation ModelBrowser99.02%0.99800.8804100% In-Browser
#7Streaming nanotabiclTabular Foundation ModelPyTorch/Python98.98%0.99800.8769soda-inria/nanotabicl ↗
#8Full TabICLv2 (PyTorch Reference)Tabular Foundation ModelPyTorch/Python98.98%0.99800.8763soda-inria/tabicl ↗
#9nanotabicl VanillaTabular Foundation ModelPyTorch/Python98.96%0.99800.8745soda-inria/nanotabicl ↗
#10TabPFN v3Tabular Foundation ModelPyTorch/Python98.96%0.99800.8703PriorLabs/tabpfn ↗
#11pipeline.Pipeline(columntransformer=sklearn.compose. column transformer.ColumnTransformer(numeric=sklearn.pipeline.Pipeline(imputer=sklearn.preprocessing.imputation.Imputer,standardscaler=sklearn.preprocessing.data.StandardScaler),nominal=sklearn.pipeline.Pipeline(simpleimputer=sklearn.impute. base.SimpleImputer,onehotencoder=sklearn.preprocessing. encoders.OneHotEncoder)),variancethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,gradientboostingclassifier=sklearn.ensemble.gradient boosting.GradientBoostingClassifier)Gradient BoostingPython98.44%0.99940.9844OpenML #10418598 ↗
#12EXAONE TabularTabular Foundation ModelPyTorch/Python98.42%0.99600.7987LGAI-Research/EXAONE-Tabular ↗
#13LogitBoost(weka.classifiers.trees.REPTree)Logistic / Linear ModelPython98.09%0.99910.9809OpenML #10416559 ↗
#14pipeline.Pipeline(imputation=preprocessing.ConditionalImputer2,catencoding=preprocessing.MultiLabelEncoder,variencethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,clf=sklearn.ensemble.forest.RandomForestClassifier)Random ForestPython98.03%0.99950.9805OpenML #9918577 ↗
#15pipeline.Pipeline(columntransformer=sklearn.compose. column transformer.ColumnTransformer(numeric=sklearn.pipeline.Pipeline(imputer=sklearn.preprocessing.imputation.Imputer,standardscaler=sklearn.preprocessing.data.StandardScaler),nominal=sklearn.pipeline.Pipeline(simpleimputer=sklearn.impute. base.SimpleImputer,onehotencoder=sklearn.preprocessing. encoders.OneHotEncoder)),variancethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,decisiontreeclassifier=sklearn.tree.tree.DecisionTreeClassifier)Decision TreePython97.92%0.98070.9791OpenML #10417868 ↗
#16pipeline.Pipeline(columntransformer=sklearn.compose. column transformer.ColumnTransformer(numeric=sklearn.pipeline.Pipeline(imputer=sklearn.preprocessing.imputation.Imputer,standardscaler=sklearn.preprocessing.data.StandardScaler),nominal=sklearn.pipeline.Pipeline(simpleimputer=sklearn.impute. base.SimpleImputer,onehotencoder=sklearn.preprocessing. encoders.OneHotEncoder)),variancethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,decisiontreeclassifier=sklearn.tree.tree.DecisionTreeClassifier)Decision TreePython97.92%0.98070.9791OpenML #10418209 ↗
#17pipeline.Pipeline(columntransformer=sklearn.compose. column transformer.ColumnTransformer(numeric=sklearn.pipeline.Pipeline(imputer=sklearn.preprocessing.imputation.Imputer,standardscaler=sklearn.preprocessing.data.StandardScaler),nominal=sklearn.pipeline.Pipeline(simpleimputer=sklearn.impute. base.SimpleImputer,onehotencoder=sklearn.preprocessing. encoders.OneHotEncoder)),variancethreshold=sklearn.feature selection.variance threshold.VarianceThreshold,adaboostclassifier=sklearn.ensemble.weight boosting.AdaBoostClassifier(base estimator=sklearn.tree.tree.DecisionTreeClassifier))AdaBoostPython97.57%0.97680.9756OpenML #10417921 ↗
STATISTICAL DISTRIBUTIONS

Variable Schema & Summary Distributions

Observed numerical ranges, category cardinalities, missing rates, and sample values across 1,728 rows.

ColVariable NameTypeMissingDistinctSummary Stats / DistributionSample Values
Acar_acceptabilityTARGETCAT044 cats: unacc, acc, good +1 moreunaccunaccunacc
Bbuying_price_tierCAT044 cats: vhigh, high, med +1 morevhighvhighvhigh
Cmaintenance_cost_tierCAT044 cats: vhigh, high, med +1 morevhighvhighvhigh
Ddoor_countCAT044 cats: 2, 3, 4 +1 more222
Epassenger_capacityCAT033 cats: 2, 4, more222
Fluggage_boot_sizeCAT033 cats: small, med, bigsmallsmallsmall
Gsafety_ratingCAT033 cats: low, med, highlowmedhigh
FREQUENTLY ASKED QUESTIONS

Frequently Asked Questions: Car

Common questions regarding the Car dataset, machine learning task formulations, and in-browser tabular inference.

What is the Car Evaluation dataset used for?

The Car Evaluation dataset is a standard benchmark used to test multi-class categorical classification, constructive induction, and decision tree algorithms by predicting overall vehicle acceptability based on six structural and economic attributes.

Can I run zero-server predictions on this dataset in Google Sheets?

Yes, Carla HQ enables zero-server in-context tabular prediction directly in spreadsheets using TabICL foundation models without needing local Python environments or backend deployments.

What machine learning models perform best on the Car Evaluation dataset?

Decision tree-based algorithms (such as Random Forest, XGBoost, and CatBoost) and tabular foundation models like TabICL excel on this dataset, consistently reaching 98% to 100% accuracy due to the clear underlying hierarchical logic.

LIVE EVALUATION IN GOOGLE SHEETS

Test TabICLv2 on Car Yourself

Open the pre-loaded Google Sheet and let Carla configure the target and task for local, zero-cloud tabular machine learning.

Step 1

Launch Carla

Click once to open the spreadsheet and Carla side panel together.

Step 2

Review the Setup

Carla selects the dataset target and task from this page automatically.

Step 3

Evaluate & Predict

Run predictions and compute metrics with zero server uploads.