Machine Learning Model for Cancer Diagnosis based on RNAseq Microarray

Abstract

The main purpose of FeRCO function is to define the minimum number of features using the most fitting reduction technique along with classification technique that give the highest classification accuracy. These techniques include Support Vector Machine (SVM) both linear and kernel, Decision Trees (DT), Random Forest (RF), K-Nearest-Neighbours (KNN) and Naïve Bayes (NB). Principle Component Analysis (PCA) both linear and kernel, LinearDiscriminant Analysis (LDA) and Factor Analysis (FA) along with different machine learning techniques were used to find a lower-dimensional subspace with better discriminatory features for better classification. The major outcomes of this research can be considered as a roadmap for interesting researchers in this field to be able to choose the most suitable machine learning algorithm whatever classification or reduction. The results show that FA and LPCA are the best reduction techniques to be used with the three datasets providing an accuracy up to 100% with TCGA and simulation datasets and accuracy up to 97.86% with WDBC datasets. LSVM is the best classification technique to be used with Linear PCA (LPCA), FA and LDA. RF is the best classification technique to be used with Kernel PCA (KPCA).

A novel deep autoencoder based survival analysis approach for microarray dataset

“…We compare our proposed autoencoder feature selection technique for survival analysis with the widely used feature selection algorithm, Principal Component Analysis (PCA) (Torkey et al, 2020) and Feature Importance (FI) (Wang & Li, 2017b). Then using these selected feature sets for survival analysis and computing Co-index and p-value for KME.…”

“…Other techniques, like Linear Discriminant Analysis (LDA), Non-negative Matrix Factorization (NMF), and Singular Value Decomposition (SVD) are also often used for dimensionality reduction (Van Der Maaten, Postma & den Herik, 2009). Dimensionality reduction methods (Torkey et al, 2020) transform a high-dimensional space into a low-dimensional space of data, producing a new representation of features. The new representation of features is considered a problem for survival analysis methods because it eliminates the property of predicting the importance of original features (genes) as it produces a new representation of features.…”