How mittrayash/parkinson-s-disease-detection-using-gait-analysis works
This repository is a notebook-based Parkinson’s disease gait-analysis and classification project. Its main workflow compresses raw gait sensor measurements into statistical features, combines patient and control data, then trains classifiers to distinguish the groups. The README also documents PCA experiments and several visualization and time-series analyses. The tree and README support these product stages, but sampled source is limited, so links between individual notebooks and particular output files are not asserted where implementation details are unavailable.
Main entry points: Combined Datasets Patient/control sets · Transformed Data Transformed features · merge_datasets.ipynb Statistical compression · CSV Gait CSVs
Components 16
Gait data
- Gait CSVs
CSV - Transformed features
Transformed Data - Patient/control sets
Combined Datasets - PCA features
PCA
Feature analysis
- Statistical compression
merge_datasets.ipynb - Dataset preparation
merge_datasets.ipynb - PCA experiments
pca_without_poly.ipynb
Classification
- Logistic regression
merge_datasets_Logistic.ipynb - Decision trees
merged_datasets_decision trees-checkpoint.ipynb - Random forests
merge_datasets_random_forests.ipynb - SVM kernels
merge_datasets_SVM.ipynb - k-nearest neighbors
merge_datasets_KNN.ipynb
Results and visuals
- Feature visualization
Most_Prominent_Visual.ipynb - Component analysis
Significant_Components_2d.ipynb - Time-series analysis
Time Series Analysis.ipynb
Other components
- Analyst
Connections 15
- Analyst → Statistical compression: runs analysis
- Gait CSVs → Statistical compression: reads gait data
- Statistical compression → Transformed features: writes features
- Transformed features → Dataset preparation: reads features
- Dataset preparation → Patient/control sets: writes datasets
- Patient/control sets → Logistic regression: trains on
- Patient/control sets → Decision trees: trains on
- Patient/control sets → Random forests: trains on
- Patient/control sets → SVM kernels: trains on
- Patient/control sets → k-nearest neighbors: trains on
- Transformed features → PCA experiments: reads features
- PCA experiments → PCA features: writes components
- Patient/control sets → Feature visualization: analyzes data
- PCA features → Component analysis: analyzes components
- Gait CSVs → Time-series analysis: analyzes gait