KPCA-LDA

KPCA+LDA project for Statistical ML Lesson.
Core folder includes simple implementations of KPCA and LDA from scratch.
KPCA+LDA is being applied for seizure detection and cancer recognition from rna seq.
This project produces results, POV vs number of features diagrams for KPCA and LDA, Data visualisation in 1D, 2D and 3D and classification examples for EEGs.

Usage

Install requirements

pip install -r path_to_KPCA+LDA_project/requirements.txt

Download datasets

1. Seizure Detection
2. RNA Seq

Add files to project

• Create a folder on root ex. datasets and move downloaded datasets there.
• Open definitions.py from root.
• Set DATASETS variable with the datasets folder name ex. DATASETS='datasets'.
• Set DATASET_SEIZURE with the filename of the seizures dataset ex. DATASET_SEIZURE='seizure.csv'.
• Set DATA_GENES with the filename of the RNA Seq data ex. DATA_GENES='genes.csv'.
• Set LABELS_GENES with the filename conaining the classes for RNA Seq dataset ex. LABELS_GENES='genes_labels.csv'.
• Extra configurations can be optionaly made through the definitions.py file.

Examples

Log

----------- 2018-12-23 23:24:09.387296 -----------

Loading Dataset...
6900 training data loaded
4600 test data loaded
Preprocessing...
	Scaling data, using Min Max Scaler with params:
	{'copy': True, 'feature_range': (-1, 1)}
	Applying Principal Component Analysis with params:
	{'kernel': 'RBF', 'alpha': 'auto', 'coefficient': 0, 'degree': 3, 'sigma': 0.3, 'n_components': 0.95}
	Applying Linear Discriminant Analysis with params:
	{'n_components': 'auto', 'remove_zeros': True}
Creating KNN model with params:
{'algorithm': 'auto', 'leaf_size': 30, 'metric': 'minkowski', 'metric_params': None, 'n_jobs': None, 'n_neighbors': 50, 'p': 2, 'weights': 'uniform'}
Fitting model...
Model has been fit in 0.0334 seconds.
Predicting...
Prediction finished in 0.398 seconds.
Model's Results:
Epileptic Seizure Accuracy: 0.612
Tumor Located Accuracy: 0.6761
Healthy Area Accuracy: 0.3641
Eyes Closed Accuracy: 0.375
Eyes Opened Accuracy: 0.9478
Accuracy: 0.595
Precision: 0.5863
Recall: 0.595
F1: 0.5869

Images

KPCA on Genes Dataset.

LDA on Genes Dataset, after KPCA.

LDA on EEG Dataset

POV vs K with Linear KPCA (same result as PCA) on Genes Dataset

EEG of eyes closed example.

EEG of epileptic seizure example.