Skip to main content
CARLA HQ
Telecommunications & Subscription Services Commerce & Industry use casesNACE J61Binary ClassificationTarget: churn_statusOpenML #42178 ↗

Telco Customer Churn

Benchmark telecommunications dataset comprising 7,043 subscriber profiles designed to predict customer churn based on demographics, subscribed services, account tenure, and billing details.

✨ Try in CarlaInstall Chrome Extension ↗100% In-Browser WebGPU • Zero Cloud Upload
7,043Records (Rows)
20Predictive Features
4 / 17Numeric / Categorical
0.0%Missing Value Ratio
0.14881 - AUC Error
10.24sWebGPU Mean Split
BUSINESS CONTEXT & OBJECTIVETelecommunications & Digital Subscription Services (J61)

Business Objective: Telco Customer Churn

01Business Context

Subscription telecommunication providers operate in highly competitive, commoditized markets where acquiring a new customer costs 5 to 7 times more than retaining an existing subscriber. Identifying dissatisfaction and flight risk before service cancellation is vital for recurring revenue stability.

02Analytical Objective

Predict the binary probability of an individual customer terminating their subscription within the upcoming month based on usage patterns, contract terms, and payment behavior.

03Economic & Decision Impact

Accurate churn prediction allows proactive retention outreach and targeted discount incentives to protect customer lifetime value (LTV). A false negative results in permanent churn and lost monthly recurring revenue, whereas a false positive incurs unnecessary retention discount costs without retention upside.

ML Benchmark Narrative

Standard GBDT implementations (XGBoost, LightGBM, and CatBoost) achieve strong benchmark performance on this dataset with ROC-AUC scores typically ranging between 0.84 and 0.86 when properly encoding categorical interactions and tenure. Modern tabular foundation models such as TabICL (Carla) achieve competitive zero-shot classification performance directly within in-context inference without requiring feature engineering or hyperparameter tuning pipelines.

Source Origin:IBM Cognos Analytics Community / Kaggle / OpenML
Creator:IBM Corporation (2018)
License:CC0: Public Domain
FEATURE SPECIFICATION & DATA DICTIONARY

Target & Feature Column Definitions

Exhaustive business definitions, measurement units, roles, and target variables across all 21 columns.

ColVariable / Feature NameRoleTypeUnit / ScaleDescription & Business Meaning
Achurn_statusChurn StatusTARGETCAT
Target variable indicating whether the customer terminated their service contract within the last month (Yes, No).
Bcustomer_idCustomer IDFeatureCAT
Unique alphanumeric identifier assigned to each telecom subscriber.
CgenderGenderFeatureCAT
Biological sex/gender classification of the customer (Female, Male).
Dis_senior_citizenSenior Citizen StatusFeatureNUM
Binary indicator specifying whether the customer is a senior citizen (1 = Yes, 0 = No).
Ehas_partnerPartner StatusFeatureCAT
Indicator of whether the customer has a domestic partner or spouse (Yes, No).
Fhas_dependentsDependents StatusFeatureCAT
Indicator of whether the customer lives with dependents such as children or elderly family members (Yes, No).
Gtenure_monthsTenureFeatureNUMmonths
Total number of consecutive months the customer has maintained an active subscription with the company.
Hhas_phone_servicePhone ServiceFeatureCAT
Indicates whether the customer subscribes to standard landline phone service (Yes, No).
Imultiple_lines_statusMultiple LinesFeatureCAT
Indicates whether the customer has multiple phone lines (Yes, No, No phone service).
Jinternet_service_typeInternet Service ProviderFeatureCAT
The category of broadband internet connection utilized by the customer (DSL, Fiber optic, No).
Konline_security_addonOnline SecurityFeatureCAT
Indicates if the customer subscribes to add-on online threat security features (Yes, No, No internet service).
Lonline_backup_addonOnline BackupFeatureCAT
Indicates if the customer subscribes to cloud data backup services (Yes, No, No internet service).
Mdevice_protection_planDevice ProtectionFeatureCAT
Indicates if the customer has subscribed to hardware warranty and insurance protection (Yes, No, No internet service).
Ntech_support_addonTechnical SupportFeatureCAT
Indicates if the customer has premium technical support included in their subscription (Yes, No, No internet service).
Ostreaming_tv_addonStreaming TVFeatureCAT
Indicates whether the subscriber uses third-party streaming television services provided over broadband (Yes, No, No internet service).
Pstreaming_movies_addonStreaming MoviesFeatureCAT
Indicates whether the subscriber streams on-demand movies via provider packages (Yes, No, No internet service).
Qcontract_termContract TermFeatureCAT
Customer agreement commitment duration (Month-to-month, One year, Two year).
Rpaperless_billing_enabledPaperless BillingFeatureCAT
Indicates whether invoice statements are delivered electronically rather than via postal mail (Yes, No).
Spayment_methodPayment MethodFeatureCAT
Customer's selected invoice settlement option (Electronic check, Mailed check, Bank transfer (automatic), Credit card (automatic)).
Tmonthly_charges_usdMonthly ChargesFeatureNUMUSD
The recurring bill amount charged to the subscriber each month.
Utotal_charges_usdTotal ChargesFeatureNUMUSD
Cumulative gross amount billed to the customer across their entire account lifecycle.
DATASET PREVIEW

Interactive Data Table

Explore rows, feature values, and target labels for Telco Customer Churn.

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 →
10.24s9 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 →
92.17sIncludes warmup & sync
Evaluation Protocol?Precomputed row indices ensure the browser and Python runners evaluate identical out-of-sample observations without leakage.Explore Split Protocol →
3×3 SplitsRepeated IID • 9 total
TabICLv2 Score?Primary task performance score (1 - AUC Error) achieved by TabICLv2 across out-of-sample evaluation splits.Explore Metric Formula →
0.14881 - AUC Error
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 / F14,6952,34810.44s81.09%0.85310.60710.67650.5506
R1 / F24,6952,34810.31s80.15%0.85160.57640.66460.5088
R1 / F34,6962,34710.10s80.36%0.85020.58050.67020.5120
R2 / F14,6952,34810.33s80.79%0.85650.59190.67840.5249
R2 / F24,6952,34810.29s80.15%0.85170.58470.65730.5265
R2 / F34,6962,34710.11s80.74%0.84670.58680.68150.5152
R3 / F14,6952,34810.38s82.41%0.87240.61940.72730.5393
R3 / F24,6952,34810.24s79.43%0.84810.56840.64110.5104
R3 / F34,6962,3479.95s79.76%0.83100.57930.64620.5249
Mean ± Std10.24s80.54% ± 0.82%0.851 ± 0.0100.588 ± 0.0150.6710.524
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
#1TabPFN v3Tabular Foundation ModelPyTorch/Python80.59%0.85140.5864PriorLabs/tabpfn ↗
#2EXAONE TabularTabular Foundation ModelPyTorch/Python80.56%0.85160.5886LGAI-Research/EXAONE-Tabular ↗
#3Full TabICLv2 (PyTorch Reference)Tabular Foundation ModelPyTorch/Python80.55%0.85120.5884soda-inria/tabicl ↗
#4Google TabFM v1.0Tabular Foundation ModelPyTorch/Python80.55%0.85260.5864google-research/tabfm ↗
#5Carla Engine (TabICLv2 WebGPU)Tabular Foundation ModelBrowser80.54%0.85130.5883100% In-Browser
#6nanotabicl VanillaTabular Foundation ModelPyTorch/Python80.53%0.85120.5878soda-inria/nanotabicl ↗
#7Streaming nanotabiclTabular Foundation ModelPyTorch/Python80.49%0.85110.5868soda-inria/nanotabicl ↗
STATISTICAL DISTRIBUTIONS

Variable Schema & Summary Distributions

Observed numerical ranges, category cardinalities, missing rates, and sample values across 7,043 rows.

ColVariable NameTypeMissingDistinctSummary Stats / DistributionSample Values
Achurn_statusTARGETCAT022 cats: No, YesNoNoYes
Bcustomer_idCAT07,0437590-VHVEG5575-GNVDE3668-QPYBK
CgenderCAT022 cats: Male, FemaleFemaleMaleMale
Dis_senior_citizenNUM02[0, 1] μ=0.2 σ=0.4000
Ehas_partnerCAT022 cats: No, YesYesNoNo
Fhas_dependentsCAT022 cats: No, YesNoNoNo
Gtenure_monthsNUM073[0, 72] μ=32.4 σ=24.61342
Hhas_phone_serviceCAT022 cats: Yes, NoNoYesYes
Imultiple_lines_statusCAT033 cats: No, Yes, No phone serviceNo phone serviceNoNo
Jinternet_service_typeCAT033 cats: Fiber optic, DSL, NoDSLDSLDSL
Konline_security_addonCAT033 cats: No, Yes, No internet serviceNoYesYes
Lonline_backup_addonCAT033 cats: No, Yes, No internet serviceYesNoYes
Mdevice_protection_planCAT033 cats: No, Yes, No internet serviceNoYesNo
Ntech_support_addonCAT033 cats: No, Yes, No internet serviceNoNoNo
Ostreaming_tv_addonCAT033 cats: No, Yes, No internet serviceNoNoNo
Pstreaming_movies_addonCAT033 cats: No, Yes, No internet serviceNoNoNo
Qcontract_termCAT033 cats: Month-to-month, Two year, One yearMonth-to-monthOne yearMonth-to-month
Rpaperless_billing_enabledCAT022 cats: Yes, NoYesNoYes
Spayment_methodCAT044 cats: Electronic check, Mailed check, Bank transfer (automatic) +1 moreElectronic checkMailed checkMailed check
Tmonthly_charges_usdNUM01,585[18.25, 118.75] μ=64.8 σ=30.129.8556.9553.85
Utotal_charges_usdNUM11 (0.16%)6,530[18.8, 8684.8] μ=2283.3 σ=2266.629.851889.5108.15
FREQUENTLY ASKED QUESTIONS

Frequently Asked Questions: Telco Customer Churn

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

What is the Telco Customer Churn dataset used for?

The Telco Customer Churn dataset contains 7,043 customer records used to benchmark supervised classification algorithms for predicting whether telecommunications subscribers will cancel their services.

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 building or maintaining complex Python pipelines.

What machine learning models perform best on Telco Customer Churn?

Gradient boosting frameworks like LightGBM, CatBoost, and XGBoost typically achieve ROC-AUC scores around 0.84 to 0.86, while tabular foundation models like TabICL deliver comparable accuracy zero-shot.

LIVE EVALUATION IN GOOGLE SHEETS

Test TabICLv2 on Telco Customer Churn 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.