Keywords : Random Forest Classifier
Performance Study of ML Models and Neural Networks for Detection of Parkinson Disease using Dysarthria Symptoms
European Journal of Molecular & Clinical Medicine,
2021, Volume 8, Issue 3, Pages 767-779
Parkinson Disease (PD) is brain disorder that affects the central nervous system
that results in damage of nerve cells causing dopamine to drop. PD has a severe effect on
vocal features termed as Dysarthria symptoms including varied pitch, extended pauses,
monotonous and speaking slowly or with a slur. In this paper, a dataset containing various
vocal features are taken as input to analyze the performance of various Machine Learning
algorithms including Naive Bayes, Random Forest Classifier, Support Vector Machines
(SVM), Linear Regression, K Nearest Neighbor (KNN) and Neural Networks such as ANN
and LSTM. The best classification accuracy was obtained by ANN around 90.00%.
Hybrid of PSO-GSA based Clustering Approach for Predicting Structural Class Prediction using Random Forest Method
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 10, Pages 17-32
Protein can be classified in different classes. A lot of research is being performed for analyzing the structure and classes of protein. There are many problems associated with protein structure. Some of them are folding problem and protein structure prediction (PSP) etc. PSP is the most considerable open problem in field of biology. In the present work different algorithms like particle swarm optimization (PSO), gravitational search algorithm (GSA) and K-Mean clustering algorithms are used to classify different structures of protein. A random forest (RF) classifier is proposed for analyzing and comparing different protein classes in terms of other conventionally available algorithms in terms of various performance parameters like accuracy, recall, precision and specificity. The proposed classifier proved better than other classifiers in terms of accuracy and can be helpful in predicting the protein structures. A hybrid PSO-GSA algorithm is also proposed which provided improved performance as compared to single algorithms and can be utilized for analysis of protein structure.
TONGUE IMAGE CLASSIFICATION USING RANDOM FOREST CLASSIFIER
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 9, Pages 1435-1438
DOI:
10.31838/ejmcm.07.09.152
Diabetes people who take antibiotics regularly to treat multiple infections are more likely to develop a fungal mouth and language infection. The fungus thrives in people with uncontrolled diabetes at high levels of glucose in the saliva. A yeast infection called oral thrush is common among people with diabetes. It looks like a white layer coating your language and your cheeks' insides. The Yeast grows in a higher amount of sugar found in your saliva. The early diagnosis is required for tongue image classification. In this study, the automatic classification of tongue image classification for diabetes detection system is discussed. Initially, the input tongue images are given to median filter for pre-processing. Then the Gray Level Co-occurrence Matrix (GLCM) and Haralick features are extracted. Finally, Random Forest (RF) classifier is used for Prediction. The performance of proposed system produces the classification accuracy of 95%using RF classifier.
MULTIWAVELET TRANSFORM BASED GLAUCOMA CLASSIFICATION USING RANDOM FOREST
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 9, Pages 1470-1475
DOI:
10.31838/ejmcm.07.09.158
Glaucoma is a group of eye disorders that damage the optic nerve, essential for good vision. Often this damage comes from an abnormally high pressure in the eye. The early diagnosis of glaucoma detection is required because it leads to loss of vision. The fundus images are decomposed by Multi Wavelet Transform (MWT). Then the sub-band coefficients of MWT are extracted by using energy features. Then the redundant features are reduced by Principal Component Analysis (PCA). Finally, Random Forest (RF) classifier is used for prediction. The classification results are obtained in the experimental results and discussion section. The system produces classification accuracy of 93% by using MWT based PCA reduction and RF classifier.