Browsing by Author "Keya, Maria Sultana"

Now showing 1 - 10 of 10

Comparison of Different Machine Learning Algorithm for Detecting Bankruptcy
(Daffodil International University, 2021-01-15) Keya, Maria Sultana; Akter, Himu; Sozib, Md. Atiqur Rahman
There have been severe experiments from academics and merchandisers concerning models for Predicting bankruptcy. The paper propounds an extensive rethink of work done during 5 years in the petition of intellectual strategy to accomplish bankruptcy prediction problems. Several machine learning directions are being used in this paper for Predicting bankruptcy. Some algorithms: AdaBoost, Decision tree, J48, Bagging, Random Forest are used in this paper. By traditional models, machine learning models offer enhancing bankruptcy prediction accuracy. Different types of models are tested using several evaluation metrics. The five years Bagging accuracy range is 95% within 97% among another model. Here we include k-fold cross-validation(k=10) to measure our accuracy. Bagging accuracy is high in this paper. A confusion matrix is used to recount the perfection of a classification model that gives true values for knowing.
Comparison of Different Machine Learning Algorithms for Detecting Bankruptcy
(2021 6th International Conference on Inventive Computation Technologies (ICICT), IEEE, 2021-01) Keya, Maria Sultana; Akter, Himu; Rahman, Md. Atiqur; Rahman, Md. Mahbobur; Emon, Minhaz Uddin; Zulfiker, Md. Sabab
There has been severe experiments from academics and merchandisers concerning models for Predicting bankruptcy. The paper propounds an extensive rethink of work done during 5 years in the petition of intellectual strategy to accomplish bankruptcy prediction problems. Several machine learning directions are being used in this research paper for Predicting bankruptcy. Some algorithms: AdaBoost, Decision tree, J48, Bagging, Random Forest are used in this paper. By traditional models, machine learning models offer enhancing bankruptcy prediction accuracy. Different types of models are tested using several evaluation metrics. The five years Bagging accuracy range is 95% within 97% among another model. Here include kfold cross-validation(k=10) to measure our accuracy. Bagging accuracy is high in this paper. Confusion matrix is used to recount the perfection of a classification model that gives true values for knowing.
Comparison of Different Machine Learning Algorithms for Detecting Bankruptcy
(2021 6th International Conference on Inventive Computation Technologies (ICICT), IEEE, 2021-02-26) Keya, Maria Sultana; Akter, Himu; Rahman, Md. Atiqur; Rahman, Md. Mahbobur; Emon, Minhaz Uddin
There has been severe experiments from academics and merchandisers concerning models for Predicting bankruptcy. The paper propounds an extensive rethink of work done during 5 years in the petition of intellectual strategy to accomplish bankruptcy prediction problems. Several machine learning directions are being used in this research paper for Predicting bankruptcy. Some algorithms: AdaBoost, Decision tree, J48, Bagging, Random Forest are used in this paper. By traditional models, machine learning models offer enhancing bankruptcy prediction accuracy. Different types of models are tested using several evaluation metrics. The five years Bagging accuracy range is 95% within 97% among another model. Here include kfold cross-validation (k=10) to measure our accuracy. Bagging accuracy is high in this paper. Confusion matrix is used to recount the perfection of a classification model that gives true values for knowing.
Machine Learning Based Image Classification of Papaya Disease Recognition
(2020 4th International Conference on Electronics, Communication and Aerospace Technology (ICECA), IEEE, 2020-12-28) Islam, Md. Ashiqul; Islam, Md. Shahriar; Hossen, Md. Sagar; Emon, Minhaz Uddin; Keya, Maria Sultana; Habib, Ahsan
To help farmers and rural people of Bangladesh, many research works are proposed in the recent years to recognize the papaya diseases that takes a great deal of advantage in machine learning fields. This research is mainly required to support agriculture to make it highly effective and helpful particularly for papaya cultivation. The primary objective of this paper is to compare some algorithms for papaya disease recognition and identify the ailment by capturing image and classify them based on their diseases with an intelligent system. To overcome this advantage, the recognition of papaya diseases will mainly involve two challenges and those are detecting the disease and classifying the diseases based on their symptoms. The proposed system is presenting an online machine learning based papaya disease in which a person captures an image via mobile app and sends it to the system for disease detection and also compare some algorithms accuracy those are random forest, k-means clustering, SVC and CNN. The system process the images and will give feedback. This intelligent system can easily detect the diseases with a high accuracy of about 98.4% to predict the papaya diseases.
Measuring the Heart Attack Possibility using Different Types of Machine Learning Algorithms
(International Conference on Artificial Intelligence and Smart Systems (ICAIS), IEEE, 2021-04-12) Keya, Maria Sultana; Shamsojjaman, Muhammad; Hossain, Faruq; Akter, Farzana; Islam, Fakrul; Emon, Minhaz Uddin
The heart seems to be a very complicated organ in human body. If some part of the heart has been seriously damaged, the remaining part of the heart will still remain functioning. But as a result of the injury, the heart can be weakened and unable to pump as much blood as normal. With timely detection of multiple possible hamstring issues, proper care, and dietary changes after a heart attack, the additional injury can be reduced or avoided. In this paper, different types of machine learning algorithms are used for measuring the possibility heart attack, they are logistic regression, random forest, bagging, MLP, and decision tree. By finding the best algorithm, this paper also shows the correlation matrices, visualizes the feature, and AUC. From this research work, it is evident that the logistic regression is the best model with an accuracy of about 80% and also gives the best AUC of about 87%.
Performance Analysis of Chronic Kidney Disease through Machine Learning Approaches
(Scopus, 2021) Emon, Minhaz Uddin; Imran, Al Mahmud; Islam, Rakibul; Keya, Maria Sultana; Zannat, Raihana; Ohidujjaman, Ohidujjaman
Data mining and machine learning play a vital role in health care and also medical information and detection, Now a day machine learning techniques use awareness of some major health risks such as diabetic prediction, brain tumor detection, covid 19 detections, and many more. The kidney is the most important organ of our body and if it has any problem then the impact is more dangerous to our body. Chronic kidney disease (CKD), otherwise referred to as renal disease. CKD requires disorders that damage and reduce the capacity of our kidneys to keep us healthy. So, it is required to be concerned about kidney disease to our very primary stage. We take a few attributes to measure our analysis about chronic kidney disease and this attribute is one of the major occurrences of chronic kidney disease. Therefore 8 machine learning classifier are used to measure analysis using weka tools namely: Logistic Regression (LG), Naive Bayes (NB), Multilayer Perceptron (MLP), Stochastic Gradient Descent (SGD), Adaptive Boosting (Adaboost), Bagging, Decision Tree (DT), Random Forest (RF) classifier are used. We feature extraction of all attributes using principal component analysis (PCA). We gain the highest accuracy from the Random Forest (RF) and it is 99 % and ROC (receiver operating characteristic) curve value is also highest from other algorithms.
Performance Analysis of Diabetic Retinopathy Prediction Using Machine Learning Models
(2021 6th International Conference on Inventive Computation Technologies (ICICT), IEEE, 2021-02-26) Emon, Minhaz Uddin; Zannat, Raihana; Khatun, Tania; Rahman, Mahfujur; Keya, Maria Sultana
Diabetic Retinopathy (DR) is a symptom of diabetes that affects the eyes. The blood vessels of the light tissue behind the eyes are damaged (retina). Machine Learning (ML) techniques play a vital role in computer aid diagnosis and discover successful systems for detecting life-threatening diseases. This research aimed to predict diabetic retinopathy and also implement feature extraction to figure out some features. In this research, the data is collected from the UCI machine learning repository. Several Machine Learning (ML) techniques are used for analysis this dataset and find out the best performance and sensitivity, selectivity, true positive (tp) rate, false negative (fn) rate and receiver operating characteristic (roc) curve. In this study, some machine learning algorithms are used such as Naive Bayes, Sequential Minimal Optimization (SMO), logistic regression, Stochastic Gradient Descent (SGD), bagging classifier, J48 classifier, decision tree classifier, and random forest classifier. The overall performance of logistic regression shows the best result.
Performance Analysis of Machine Learning Approaches in Stroke Prediction
(Scopus, 2020-12-28) Emon, Minhaz Uddin; Keya, Maria Sultana; Meghla, Tamara Islam; Rahman, Md. Mahfujur; Al Mamun; M Shamim; Kaiser, M Shamim
Most of strokes will occur due to an unexpected obstruction of courses by prompting both the brain and heart. Early awareness for different warning signs of stroke can minimize the stroke. This research work proposes an early prediction of stroke diseases by using different machine learning approaches with the occurrence of hypertension, body mass index level, heart disease, average glucose level, smoking status, previous stroke and age. Using these high features attributes, ten different classifiers have been trained, they are Logistics Regression, Stochastic Gradient Descent, Decision Tree Classifier, AdaBoost Classifier, Gaussian Classifier, Quadratic Discriminant Analysis, Multi layer Perceptron Classifier, KNeighbors Classifier, Gradient Boosting Classifier, and XGBoost Classifier for predicting the stroke. Afterwards, results of the base classifiers are aggregated by using the weighted voting approach to reach highest accuracy. Moreover, the proposed study has achieved an accuracy of 97%, where the weighted voting classifier performs better than the base classifiers. This model gives the best accuracy for the stroke prediction. The area under curve value of weighted voting classifier is also high. False positive rate and false negative rate of weighted classifier is lowest compared with others. As a result, weighted voting is almost the perfect classifier for predicting the stroke that can be used by physicians and patients to prescribe and early detect a potential stroke.
Predicting Performance Analysis of Garments Women Working Status in Bangladesh Using Machine Learning Approaches
(2021 6th International Conference on Inventive Computation Technologies (ICICT), IEEE, 2021-02-26) Keya, Maria Sultana; Emon, Minhaz Uddin; Akter, Himu; Imran, Md. Al Mahmud; Hassan, Md. Kamrul; Mojumdar, Mayen Uddin
Maria Sultana Keya, Minhaz Uddin Emon, Himu Akter, Md. Al Mahmud Imran, Md. Kamrul Hassan, Mayen Uddin Mojumdar Abstract: In Bangladesh, the garment industry has played an important role in economically uplifting a diverse community of poor and marginalized people. There are now 4,825 garment factories that employ more than three million people. Completely 85% of these employees are female. But most of the female workers work to support their family and also contribute his family to lead a minimum life. In this paper, we try to find out relation between their health status, their family earning, their family member information, their working time or how many year they work in this sector and how many time they want to work. The dataset is collected from the Ashulia and Gazipur area garments of Bangladesh. This research work has observed that most of the female workers work at finishing, swing, helper, and cleaner sector. In this sector they cannot get huge salary that's why their income is limited and the range of their salaries is very low. It has also been found that, some women manage their whole family with their own income. Besides they are feeling bored with the same work. Nowadays machine learning and data mining tools play a vital role in finding the measurement of some important factors. This paper analyses the women working performance based on their previous activity and use some machine learning algorithms likely: Decision Tree Classifier (DTC), Logistic Regression (LR), Random Forest Classifier (RFC), and Stochastic Gradient Descent (SGD) we get the best result from Logistic Regression (LR) and it is 69%.
Primary Stage of Diabetes Prediction Using Machine Learning Approaches
(International Conference on Artificial Intelligence and Smart Systems (ICAIS), IEEE, 2021-04-12) Emon, Minhaz Uddin; Keya, Maria Sultana; Kaiser, Md. Salman; islam, Md. Ariful; Tanha, Tabassum; Zulfiker, Md. Sabab
As per the report of the World Health Organization (WHO), diabetes has become one of the rapidly expanding chronic diseases that has affected the life of 422 million people all over the world. The number of deaths in Bangladesh due to diabetes has reached 28,065, which is 3.61% of the total deaths of Bangladesh, according to the latest data published by the WHO in 2018. So we need to be concerned about the risks of diabetes disease. If we cannot take proper steps to diagnose diabetes at an early stage, eventually we have to face serious health issues. In this paper, we have shown the relation of different symptoms and diseases that cause diabetes so that we can help a person to diagnose diabetes at an early stage. Nowadays, machine learning classification approaches are well accepted by researchers for developing disease risk prediction models. Therefore eleven machine learning classification algorithms such as Logistic Regression (LR), Gaussian Process (GP), Adaptive Boosting (AdaBoost), Decision Tree (DT), K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), Support Vector Machine (SVM), Bernoulli Naive Bayes (BNB), Bagging Classifier (BC), Random Forest (RF), and Quadratic Discriminant Analysis (QDA) have been used in this study. Among all these machine learning classifiers, Random Forest (RF) classifier has showed the best accuracy of 98%. And its Area Under Curve(AUC) is also the highest.