kurzti88066/Fahrsimulator_MSY2526_AI
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

- added accuracy results for xgboost models

Browse Source
This commit is contained in:
TimoKurz 2026-03-19 17:37:32 +01:00
parent 145a5ecf78
commit 3701d11c77
1 changed files with 18 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
project_report.md
View File

@ -132,9 +132,9 @@ To establish a performance baseline, a classical Extreme Gradient Boosting (XGBo
| Metric / Model | Classical XGBoost | | Metric / Model | Classical XGBoost |
| --- | --- | | --- | --- |
| Accuracy | | | Accuracy | 0.581 |
| AUC | | | AUC | 0.562 |
| F1-Score | | | F1-Score | 0.652 |
### 4.2.2 XGBoost with GroupKFold Validation ### 4.2.2 XGBoost with GroupKFold Validation
@ -142,9 +142,9 @@ To address the challenge of inter-subject variability, the validation strategy w
| Metric / Model | XGBoost (GroupKFold) | | Metric / Model | XGBoost (GroupKFold) |
| --- | --- | | --- | --- |
| Accuracy | | | Accuracy | 0.586 |
| AUC | | | AUC | 0.573 |
| F1-Score | | | F1-Score | 0.651 |
### 4.2.3 Hybrid XGBoost with Autoencoder ### 4.2.3 Hybrid XGBoost with Autoencoder
@ -152,9 +152,9 @@ To improve feature quality, a hybrid approach was introduced by pre-training a d
| Metric / Model | XGBoost + Autoencoder | | Metric / Model | XGBoost + Autoencoder |
| --- | --- | | --- | --- |
| Accuracy | | | Accuracy | 0.589 |
| AUC | | | AUC | 0.575 |
| F1-Score | | | F1-Score | 0.650 |
### 4.2.4 Robust XGBoost with MAD Outlier Removal ### 4.2.4 Robust XGBoost with MAD Outlier Removal
@ -162,9 +162,9 @@ Recognizing that physiological and AU data often contain sensor artifacts, a rob
| Metric / Model | XGBoost + MAD | | Metric / Model | XGBoost + MAD |
| --- | --- | | --- | --- |
| Accuracy | | | Accuracy | 0.641 |
| AUC | | | AUC | 0.610 |
| F1-Score | | | F1-Score | 0.733 |
### 4.2.5 Combined Dataset of Action Units and EyeTracking ### 4.2.5 Combined Dataset of Action Units and EyeTracking
@ -174,9 +174,9 @@ By applying performance-based subject splitting, we ensured that the training an
| Metric / Model | Final Combined Model | | Metric / Model | Final Combined Model |
| --- | --- | | --- | --- |
| Accuracy | | | Accuracy | 0.659 |
| AUC | | | AUC | 0.621 |
| F1-Score | | | F1-Score | 0.715 |
### 4.2.6 Regularized XGBoost with Complexity Control ### 4.2.6 Regularized XGBoost with Complexity Control
@ -186,9 +186,9 @@ By penalizing large weights and promoting feature sparsity, the model is forced
| Metric / Model | Regularized XGBoost | | Metric / Model | Regularized XGBoost |
| --- | --- | | --- | --- |
| Accuracy | | | Accuracy | 0.665 |
| AUC | | | AUC | 0.646 |
| F1-Score | | | F1-Score | 0.727 |
### 4.3 Isolation Forest ### 4.3 Isolation Forest
To start with unsupervised learning techniques, `IsolationForest.ipynb`was created to research how well a simple ensemble classificator performs on the created dataset. To start with unsupervised learning techniques, `IsolationForest.ipynb`was created to research how well a simple ensemble classificator performs on the created dataset.