{"id":31556923,"url":"https://github.com/rachkat/predictive-modeling-customer-targeting","last_synced_at":"2026-07-18T16:05:11.731Z","repository":{"id":316474103,"uuid":"1063543691","full_name":"rachkat/predictive-modeling-customer-targeting","owner":"rachkat","description":"Predictive analytics project (Logistic Regression + Random Forest) using CoIL Challenge 2000 dataset to optimize customer targeting for insurance acquisition.","archived":false,"fork":false,"pushed_at":"2025-09-24T19:27:19.000Z","size":3824,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-09-24T21:25:26.920Z","etag":null,"topics":["crm-integration","data-science","machine-learning","marketing-analytics","r","random-forest"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/rachkat.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-09-24T19:16:44.000Z","updated_at":"2025-09-24T19:27:23.000Z","dependencies_parsed_at":"2025-09-24T21:35:35.558Z","dependency_job_id":null,"html_url":"https://github.com/rachkat/predictive-modeling-customer-targeting","commit_stats":null,"previous_names":["rachkat/predictive-modeling-customer-targeting"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/rachkat/predictive-modeling-customer-targeting","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rachkat%2Fpredictive-modeling-customer-targeting","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rachkat%2Fpredictive-modeling-customer-targeting/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rachkat%2Fpredictive-modeling-customer-targeting/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rachkat%2Fpredictive-modeling-customer-targeting/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/rachkat","download_url":"https://codeload.github.com/rachkat/predictive-modeling-customer-targeting/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rachkat%2Fpredictive-modeling-customer-targeting/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":278386517,"owners_count":25978187,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-04T02:00:05.491Z","response_time":63,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["crm-integration","data-science","machine-learning","marketing-analytics","r","random-forest"],"created_at":"2025-10-04T23:19:46.789Z","updated_at":"2025-10-04T23:19:51.154Z","avatar_url":"https://github.com/rachkat.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# Improving Customer Targeting Through Predictive Modeling  \n\n![Made with R](https://img.shields.io/badge/Made%20with-R-276DC3?logo=r\u0026logoColor=white\u0026style=flat-square) \n![License: MIT](https://img.shields.io/badge/License-MIT-green?style=flat-square) \n![Status: Finished](https://img.shields.io/badge/Status-Finished-brightgreen?style=flat-square)\n\n---\n\n## Executive Summary  \nThis project applies **predictive analytics** to improve customer acquisition strategies for **The Insurance Company (TIC)**, using the **CoIL Challenge 2000 dataset** (5,822 customers × 86 variables).  \n\nBy testing **Logistic Regression** and **Random Forest**, the study demonstrates how predictive modeling can:  \n- Improve **customer targeting** for mobile home insurance.  \n- Reduce **marketing acquisition costs** by prioritizing high-probability leads.  \n- Provide **data-driven insights** for long-term customer relationship management.  \n\nKey takeaway: **Logistic Regression** provided interpretability for deployment, while **Random Forest** delivered higher predictive accuracy by capturing complex interactions.  \n\n📄 **Full report PDF** → [Improving-Customer-Targeting-Through-Predictive-Modeling.pdf](./Improving-Customer-Targeting-Through-Predictive-Modeling.pdf)  \n\n---\n\n## Project Context  \n**Business Challenge**  \n- Traditional marketing = high cost + low conversion.  \n- TIC needed to identify **which customers are most likely to purchase** mobile home insurance policies.  \n\n**Solution Approach**  \n- Apply **predictive modeling** to customer demographic \u0026 product ownership data.  \n- Compare **Logistic Regression** (interpretable baseline) vs. **Random Forest** (higher complexity).  \n- Evaluate models on **accuracy, precision, recall, F1-score, ROC-AUC**.  \n\n---\n\n## Dataset  \n- **Source**: CoIL Challenge 2000 (real-world business data).  \n- **Size**: 5,822 rows × 86 columns.  \n- **Features**:  \n  - Demographic (age, household size, income, education).  \n  - Product ownership (car, life, home policies).  \n  - Purchasing power \u0026 socio-economic indicators.  \n- **Target**: `CARAVAN` → binary (owns mobile home insurance: 1 = yes, 0 = no).  \n\n---\n\n## Methods  \n\n### Data Preparation  \n- Converted categorical/factor variables.  \n- Split into 80/20 training vs. testing sets.  \n- Addressed severe **class imbalance** (only ~6% policyholders).  \n\n### Algorithms  \n1. **Logistic Regression (GLM)**  \n   - Probability-based, interpretable, fast to deploy.  \n2. **Random Forest**  \n   - Ensemble method, improved accuracy, captured nonlinear relationships.  \n\n### Evaluation  \n- Metrics: Accuracy, Precision, Recall, F1-score, ROC-AUC.  \n- Visuals: Decision trees, feature importance, confusion matrix, ROC curve.  \n\n---\n\n## Results \u0026 Insights  \n\n- **Logistic Regression**:  \n  - Strong interpretability, useful for pilot deployment.  \n  - High **precision (78.1%)**, but low **recall (7.1%)** → missed many true buyers.  \n\n- **Random Forest**:  \n  - Better handling of complex feature interactions.  \n  - **Feature importance**: Purchasing behavior (PBRAND), demographics (MOSTYPE), car/life insurance ownership (APERSAUT, PPERSAUT), and purchasing power (MKOOPKLA) drove predictions.  \n\n- **Business Value**:  \n  - Improved targeting efficiency → fewer wasted marketing efforts.  \n  - Actionable insights for segmentation \u0026 cross-selling.  \n  - Foundation for **CRM integration** and **predictive lead scoring**.  \n\n---\n\n## Key Skills Demonstrated  \n\n- **Predictive Modeling**: Logistic Regression \u0026 Random Forest.  \n- **Data Wrangling**: Handling imbalanced data, feature engineering, correlation analysis.  \n- **Model Evaluation**: Confusion matrices, precision-recall tradeoffs, ROC-AUC.  \n- **Business Translation**: Linking analytics results to marketing efficiency \u0026 acquisition cost reduction.  \n- **Reproducibility**: Structured process with R Markdown, version control in GitHub.  \n\n---\n\n## Reproducibility  \n\n**Environment**: RStudio, R 4.2.1  \n**Packages**: `stats`, `randomForest`, `caret`, `ggplot2`  \n\n```r\n# Example: Logistic Regression\nglm_model \u003c- glm(CARAVAN ~ ., data = train, family = \"binomial\")\nsummary(glm_model)\n\n# Example: Random Forest\nlibrary(randomForest)\nrf_model \u003c- randomForest(CARAVAN ~ ., data = train, ntree = 500, importance = TRUE)\nvarImpPlot(rf_model)\n```\n\n---\n\n## Limitations \u0026 Next Steps  \n\n- Severe class imbalance limited recall → explore **SMOTE** or cost-sensitive learning.  \n- Add hyperparameter tuning (**grid search, Bayesian optimization**).  \n- Deploy models with **PMML** for integration into CRM systems.  \n- Pilot test with TIC’s customer data before full-scale deployment.  \n\n---\n\n## License  \nReleased under the **MIT License**. See [LICENSE](./LICENSE).  \n\n---\n\n## Tags  \n`predictive-analytics, logistic-regression, random-forest, insurance, customer-targeting, machine-learning, data-science, coil-challenge, r, marketing-analytics`  \n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frachkat%2Fpredictive-modeling-customer-targeting","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Frachkat%2Fpredictive-modeling-customer-targeting","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frachkat%2Fpredictive-modeling-customer-targeting/lists"}