Enterprise AI Administration
Telecom AI Model Registry
Full registry of telecom AI models with algorithm, features, target, performance and ownership.
Telecom AI model registry
21 governed models · production-recommended design specifications (not trained or deployed)
| Model | Algorithm | Implementation library | Model type | Explainability method | Related KPIs | Journey / mission | Governance status | Implementation status | Performance | |
|---|---|---|---|---|---|---|---|---|---|---|
ARPU Uplift ModelMDL-ARPU-UPLIFT · v2.4 | XGBoostGradient Boosted Regression Trees | XGBRegressor | Regression | SHAP TreeExplainer | Blended ARPU | Revenue & ProfitabilityRevenue & ARPU | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Capital Return ModelMDL-CAPEX-ROI · v1.1 | Deterministic financial model | — | — | — | Network investment ROI | Cost & Capital OptimizationCapital & Investment | REGISTERED | NOT IMPLEMENTED | Not yet measured | View details → |
Churn Propensity Model (legacy)MDL-CHURN-LEGACY · v2.0 | Logistic regression | — | — | — | — | Churn & RetentionChurn & Retention | RETIRED | NOT IMPLEMENTED | Not yet measured | View details → |
Churn Propensity ModelMDL-CHURN-PROP · v3.2 | XGBoostGradient Boosted Decision Trees | XGBClassifier | Binary Classification | SHAP TreeExplainer | Monthly churnRevenue at risk (churn) | Churn & RetentionChurn & Retention | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Congestion Risk ModelMDL-CONGESTION · v0.7 | Gradient Boosted Trees | — | — | — | Congestion rate | Network Performance & RiskNetwork Performance & Risk | REGISTERED | NOT IMPLEMENTED | Not yet measured | View details → |
Digital Adoption ModelMDL-DIGITAL-ADOPT · v0.4 | Logistic regression | — | — | — | — | Digital Services & ChannelsDigital Services | REGISTERED | NOT IMPLEMENTED | Not yet measured | View details → |
Energy OPEX Forecast ModelMDL-ENERGY-COST · v1.6 | SARIMAXStatistical Time-Series Forecasting | statsmodels SARIMAX | Time-Series Forecasting | Forecast decomposition / exogenous driver contribution | Energy cost | Cost & Capital OptimizationCost Optimization | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Energy Cost Driver ModelMDL-ENERGY-DRIVER · v0.1 | XGBoostGradient Boosted Regression Trees | XGBRegressor | Regression | SHAP TreeExplainer | Energy cost | Cost & Capital OptimizationCost Optimization | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Fraud Pattern Discovery ModelMDL-FRAUD-DISC · v0.1 | DBSCANDensity-Based Clustering | DBSCAN | Unsupervised Density-Based Clustering | Cluster / outlier profile | Fraud lossSIM-box incidents | Fraud & Revenue AssuranceFraud & Revenue Assurance | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Fraud & Leakage Detection ModelMDL-FRAUD-LEAK · v2.1 | Anomaly detection + rules | — | — | — | — | Fraud & Revenue AssuranceFraud & Revenue Assurance | REGISTERED | NOT IMPLEMENTED | Not yet measured | View details → |
Fraud Risk ClassifierMDL-FRAUD-RISK · v2.1 | XGBoostGradient Boosted Decision Trees | XGBClassifier | Binary / Multiclass Classification | SHAP TreeExplainer | Fraud lossSIM-box incidents | fraud-revenue-assuranceFraud & Revenue Assurance | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Revenue Leakage Detection ModelMDL-LEAK-DETECT · v1.4 | Isolation Forest + Business RulesUnsupervised Anomaly Detection + Business Rules | IsolationForest + RA rule set | Hybrid Anomaly Detection | Anomaly score + triggered rules + feature contribution where supported | Revenue leakageRevenue leakage | fraud-revenue-assuranceFraud & Revenue Assurance | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Network Incident Risk ModelMDL-NET-INCIDENT · v1.9 | XGBoostGradient Boosted Decision Trees | XGBClassifier | Classification / Incident Probability | SHAP TreeExplainer | Network availability | Network Performance & RiskNetwork Performance & Risk | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Net Adds Forecast ModelMDL-NETADDS-FCST · v1.2 | SARIMAX (benchmark: ARIMA)Statistical Time-Series Forecasting | statsmodels SARIMAX | Time-Series Forecasting | Forecast decomposition / coefficient contribution | Net adds | Subscriber GrowthSubscriber Growth | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
NPS Driver ModelMDL-NPS-DRIVER · v1.4 | Elastic Net Regression (benchmark: Multiple Linear Regression)Regularized Linear Regression | ElasticNet | Regression / Driver Analysis | Coefficient interpretation; SHAP optional where supported | NPS | Customer ExperienceCustomer Experience | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Revenue & Margin ForecastMDL-REV-FCST · v1.0 | Time series regression | — | — | — | EBITDA marginTotal revenue (QTD) | revenue-profitabilityFinancial Performance | REGISTERED | NOT IMPLEMENTED | Not yet measured | View details → |
Roaming Revenue ForecastMDL-ROAM-SENS · v0.9 | Time-series | — | — | — | Roaming revenue | Interconnect & RoamingInterconnect & Roaming | REGISTERED | NOT IMPLEMENTED | Not yet measured | View details → |
Subscriber Behavioral Segmentation ModelMDL-SEG-BEHAV · v0.1 | K-MeansCentroid-Based Clustering | KMeans | Unsupervised Clustering | Cluster profile / centroid; distance to centroid | Blended ARPUNet adds | Revenue & ProfitabilityRevenue & ARPU | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Customer Experience Segmentation ModelMDL-SEG-CX · v0.1 | K-Means (benchmark: Agglomerative Hierarchical Clustering)Clustering | KMeans; AgglomerativeClustering | Unsupervised Clustering | Cluster profile / centroid; hierarchy review | NPS | Customer ExperienceCustomer Experience | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
High-Value Subscriber Segmentation ModelMDL-SEG-HV · v0.1 | Gaussian Mixture ModelProbabilistic Clustering | GaussianMixture | Unsupervised Probabilistic Clustering | Cluster profile; membership probability | Blended ARPUMonthly churn | Revenue & ProfitabilityRevenue & ARPU | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |
Usage Pattern Segmentation ModelMDL-SEG-USAGE · v0.1 | K-MeansCentroid-Based Clustering | KMeans | Unsupervised Clustering | Cluster profile / centroid; distance to centroid | Net adds | Subscriber GrowthSubscriber Growth | APPROVED DESIGN | MODEL DEVELOPMENT REQUIRED | Not yet measured | View details → |