{"id":25296188,"url":"https://github.com/benjikazzooe/diabetes_vorhersage","last_synced_at":"2026-01-20T21:32:46.221Z","repository":{"id":274801853,"uuid":"924111120","full_name":"BenjiKazzooe/Diabetes_Vorhersage","owner":"BenjiKazzooe","description":"Machine Learning Model zur Diabetes-Klassifikation mit SHAP-Analyse ","archived":false,"fork":false,"pushed_at":"2025-01-29T13:56:42.000Z","size":19,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-06T21:16:55.033Z","etag":null,"topics":["data-science","explainable-ai","machine-learning","shap","xgboost"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/BenjiKazzooe.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"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}},"created_at":"2025-01-29T12:44:08.000Z","updated_at":"2025-01-29T13:56:46.000Z","dependencies_parsed_at":"2025-01-29T14:37:42.439Z","dependency_job_id":null,"html_url":"https://github.com/BenjiKazzooe/Diabetes_Vorhersage","commit_stats":null,"previous_names":["benjikazzooe/diabetes-classification-shap","benjikazzooe/diabetes_vorhersage"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/BenjiKazzooe/Diabetes_Vorhersage","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BenjiKazzooe%2FDiabetes_Vorhersage","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BenjiKazzooe%2FDiabetes_Vorhersage/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BenjiKazzooe%2FDiabetes_Vorhersage/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BenjiKazzooe%2FDiabetes_Vorhersage/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/BenjiKazzooe","download_url":"https://codeload.github.com/BenjiKazzooe/Diabetes_Vorhersage/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BenjiKazzooe%2FDiabetes_Vorhersage/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":28614594,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-01-20T18:56:40.769Z","status":"ssl_error","status_checked_at":"2026-01-20T18:54:26.653Z","response_time":117,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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":["data-science","explainable-ai","machine-learning","shap","xgboost"],"created_at":"2025-02-13T02:56:19.527Z","updated_at":"2026-01-20T21:32:46.195Z","avatar_url":"https://github.com/BenjiKazzooe.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"\nREADME.md:\n\n# 🏥 Diabetes-Klassifikation mit SHAP-Analyse  \n\n📊 **Maschinelles Lernen zur Vorhersage von Diabetes mit XGBoost \u0026 Explainable AI (SHAP)**  \n\n---\n\n## 🔹 **Projektbeschreibung**  \nDieses Projekt nutzt **Machine Learning (XGBoost)**, um vorherzusagen, ob eine Person an Diabetes erkrankt ist.  \nMit **SHAP (SHapley Additive Explanations)** analysieren wir die Bedeutung einzelner Features und verstehen, wie unser Modell Entscheidungen trifft.  \n\n---\n\n## 🔹 **Daten**  \n📂 **Datensatz:** [Pima Indians Diabetes Dataset](https://www.kaggle.com/datasets/nancyalaswad90/review)  \n\n**Spalten im Datensatz:**  \n- `Glucose`: Blutzucker-Wert  \n- `BMI`: Body Mass Index  \n- `Age`: Alter der Person  \n- `Pregnancies`, `BloodPressure`, `SkinThickness` u.a.  \n\n---\n\n## 🔹 **Verwendete Technologien**  \n🐍 **Python** (pandas, numpy, matplotlib)  \n🤖 **Machine Learning**: XGBoost  \n🎯 **Explainable AI**: SHAP  \n📈 **Datenvisualisierung**: Matplotlib, Seaborn  \n\n---\n\n## 🏆 Modellvergleich: XGBoost vs. Random Forest vs. Logistische Regression\n\n| Modell                     | Accuracy| Precision |   Recall   | F1-Score|\n|----------------------------|---------|-----------|------------|---------|\n| **XGBoost**                | 0.6039  | 0.4706    | **0.8727** | 0.6115  |\n| **Random Forest**          | 0.7208  | 0.6071    | 0.6182     | 0.6126  |\n| **Logi. Regression**       |**0.7532**|**0.6491**| 0.6727     |**0.6607**|\n\n### Fazit\n- **XGBoost hat den höchsten Recall** (87,3%) → Bestes (und dann auch gewähltes) Modell, wenn möglichst viele Diabetes-Fälle erkannt werden sollen.  \n- **Logistische Regression hat die höchste Accuracy (75,3%)** → Bestes Modell, wenn Balance aus Precision \u0026 Recall gewünscht ist.  \n- **Random Forest liegt in der Mitte** → Kann evtl. mit Feature Engineering verbessert werden.  \n---\n## ⏳ ARIMA-Zeitreihenanalyse für Glucose-Level\n\nDas ARIMA(5,1,0)-Modell wurde trainiert, um den Glucose-Level über die Zeit zu modellieren.\n\n**Modellstatistiken:**\n- **AIC = 7614, BIC = 7642** → Niedrigere Werte sind besser.\n- **Alle AR-Koeffizienten sind signifikant** (p-Wert \u003c 0.05).\n- **Kurtosis = 3.38, Skew = 0.20** → Fast normalverteilte Residuen.\n- **Ljung-Box-Test (`Prob(Q) = 0.58`)** → Kein Hinweis auf starke Autokorrelation.\n\n### 📌 Interpretation:\n- **Das Modell kann für Glucose-Vorhersagen genutzt werden.**\n- **Es zeigt eine autoregressive Struktur (Glucose-Level hängt von vorherigen Werten ab).**\n  \n---\n\n## 🔹 **Feature Importance Analyse mit SHAP**  \n### 📊 **Wichtigste Features laut SHAP**  \nHier eine SHAP Summary-Analyse, die zeigt, welche Features den größten Einfluss auf die Diabetes-Vorhersage haben:  \n\n- **Glucose ist der wichtigste Faktor für eine Diabetes-Diagnose**  \n- **BMI \u0026 Alter haben ebenfalls einen starken Einfluss**  \n- **Andere Faktoren (SkinThickness, BloodPressure) haben geringere Auswirkungen**  \n![Figure_1](https://github.com/user-attachments/assets/fb4df2e0-9850-4e17-b586-d63661ef5c96)\n\n---\n\n## **SHAP-Abhängigkeitsanalysen**  \n### 🔹 **Glucose vs. Diabetes-Risiko**  \n- Je höher der **Glucose-Wert**, desto größer die Wahrscheinlichkeit einer Diabetes-Erkrankung  \n- Der Effekt ist **linear** (höhere Werte = höheres Risiko)  \n\n![Figure_2](https://github.com/user-attachments/assets/3fe9f36b-0631-493d-a4c3-fa08ea2084b5)\n### 🔹 **BMI vs. Diabetes-Risiko**  \n- Ein **BMI über 30** erhöht das Diabetes-Risiko sprunghaft  \n- Ältere Personen (rote Punkte) sind stärker betroffen  \n![Figure_3](https://github.com/user-attachments/assets/4ee3611b-b677-4ef2-b23a-7ab2e414b8e5)\n\n\n\n---\n\n\n\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbenjikazzooe%2Fdiabetes_vorhersage","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbenjikazzooe%2Fdiabetes_vorhersage","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbenjikazzooe%2Fdiabetes_vorhersage/lists"}