Search

Sentiment analysis is a process of detecting sentiment through texts. It has become a very popular research work in natural language processing. Maximum researchers work with social media texts, tweets, and reviews. In this paper, we are working with Bangla text for child sentiment analysis. Child sentiment means a child's emotion which can be positive, negative, and neutral. Child sentiment analysis is needed for future development in psychology which will help researchers to detect a child's sentiment. This research paper will help to detect a child's positive, negative and neutral sentiments. To build this model we apply many machine learning and deep learning algorithms. The BERT model gives the best accuracy for our data set which is a deep learning model and MNB gives the highest accuracy from machine learning algorithms.

Polycystic Ovary Syndrome is mainly metabolic and reproductive endocrinopathy and an extremely complicated disorder among women. It causes hormonal imbalance, menstruation problems, hair loss, facial acne, facial pimple, dark skin, diabetics, overweight etc. Early detection is important for getting the curation for PCOS. Early detection brings early prevention. In our paper, we were able to collect 550 data of Bangladeshi women. Out of them, 462 are affected by this disorder. We have highlighted common symptoms of PCOS Bangladesh women and it's effect on women. We have used machine learning methods like Random Forest, Naive Bayes Classification, Decision Tree Classification, Super Vector Machine Learning, K-nearest Neighbor, Logistic Regression, XGBoost Classifier, Gradient Boosting Classifier to detect PCOS and comparing which algorithms work best for detecting the PCOS based on our real datasets. Also choose the best model on our real time dataset. We analyze our data set and find out the accuracy, confusion matrix, f1-score, cross validation, recall, precision, support. Among all classification SVM has the highest accuracy. The accuracy of the Super Vector Machine Learning is 99.09%.

Sentiment analysis is a process for mining opinion from a text. A lot of work has been done on this field for English language but the numbers don't rise high for the Bengali language. Though Bengali is the seven largest spoken language but a few works have been done for Bengali language all because of the lack of a perfect model and dataset. Everyday not only in our country but also around the world a lot of incidents happen. Because we are currently living in a global community where the news can impact a person's mind beyond borderline of countries. For this aspect it's a great opportunity to analyze the sentiment of the news. To train the model we classified the emotions into three classes. We have tried to classify a person's emotions into three different classes (happy, sad, neutral) for news published on the internet by online newspapers. For Deep learning Recurrent neural network was applied and for Machine learning, we used methods like Multinomial naive Bayes, K-nearest neighbor, Random Forest Classifier, Decision tree classifier, Support vector machine, Logistic Regression to train our model. Among those RNN obtain 95% accuracy.

Stroke is a drawn-out incapacity illness caused everywhere on the world and it is the third driving reason for demise. Pre-syndromes of stroke dives more importance now a day. Stroke happened basically because of individual way of life in the cutting the edge period changing components, for example, high glucose, coronary illness, heftiness, diabetes. In this research, we analyse different Supervised machine Learning algorithms (i.e. SVC, DT, RFC, Logistic Regression, Linear Regression, Bernoulli NB, Gaussian NB and K-Nearest Neighbours Classifier). The accuracy in the work have found highest by applying Decision Tree classifier which is 0. 93.

In this study, several aspects of the human body have been focused upon. This paper attempts to cast light on pre-and post-pathological conditions, man-machine interactions, human mindset, and ethics of AI. The paper emphasizes the cultural impacts of overeating, profuse drinking, and smoking habits. It uplifts the basic necessity of growing awareness schemes. Patients are seeking treatment in health care centers with the following serious pathological conditions and complications (We exclude the COVID-19 pandemic because it has been adequately publicized by media and press): Heart Attack, Stroke Cancer, Fatty liver & liver cirrhosis. Because of being the leading causes of sudden death prediction of heart attack is very important. Our main focus is to determine the best machine learning method. With optimal parameters, we evaluate the Dataset. Model Accuracy for the heart Attack Machine Learning Model was the highest for the Logistic Regression mode land it was 93.41%. On the contrary, the accuracy for Linear Regression Model was 60.10% which was the least.

Satisfaction is important for every situation in our life. When it is related to our future career it is more important. Our future career depends on our university life. So, university life is a very important and precious time for a student. If a student passes this life with dissatisfaction, it has an impact on their future career. There are many reasons behind dissatisfaction of students in university life. After completing college life, some student gets a chance to admitted into public university and rest of them are admitted into national university or private university. In private university they are many subjects to choose for undergraduate study. But, in recants days most of the students choose Computer Science for their undergraduate study. Some students are satisfied with their choice and some are dissatisfied. So, in our research work we are trying to find out the reason for Computer Science students' satisfaction and dissatisfaction. We collect data from Computer Science students in different private universities in Bangladesh. We found 300 participants for our research work. We collect 300 student's data and we categorize them as satisfied and dissatisfied students. In this research work we are trying to show the systematic way of collecting satisfied and dissatisfied students' data. After collecting data, we try to process, analyze and visualize the data. Therefore, our goal is to find out the reason for students' satisfaction and dissatisfaction. We make a clean dataset of those data. And Machine learning algorithms can easily classify and predict output from those data. We hope that our dataset will be helpful for the machine learning researcher. And they get appropriate output from our dataset.

Game addiction means when we are addicted to a game too much and we are forgetting our normal daily activities like sleep, eating, study etc. This is a very common problem in our generation of people. This addiction is very harmful for our physical and mental health. So, this is going to be a very dangerous fact for our generation. It mainly influences our young generation. It can change our daily life. It hampers our study concentration, brain, health and we forget our social life for this addiction. Online game addicted people face lots of problems like headache, eye problem, eating problem, lack of sleep, losing mood easily etc. Nowadays, the collection of data via online is so difficult to do. When sending data collection forms to the friends, the relatives and others they feel uncomfortable sharing their personal information. So, we can't find any standard dataset for our research. Finally, we have decided to make our own dataset via collecting data from our friends, relatives and others. And we reserve our data in a systematic way, where the ML algorithm can classify our data in some categories. Therefore, the main purpose of this research study is to make sense about online game addiction and make a good structured and large data set for researchers which can help them to analyze the data and get applicable results from their research.

An end-to-end process achieved by the convenient transformation of any purpose. Throughout this knowledge has been implemented by language advancement. In language, machine advancement happens in two ways one is text script processing, another is speech-based processing. A dual-language processing system is much needed for system development. Thus, Language processing over Bangla language is necessary for modernization because of the sub-continent most of the people use as a spoken language. We propose an identification by feature extraction from audio speech voice which is automatic speech recognition. Our model was proposed by supervised learning on continuous speech extraction over audio speech. Due to the outcome of this analysis, many fraud analyses will be possible. The purpose of model learning had used more than two hundred speakers on a voice synthesizer. All features and selected features were both identified with the frequency of male and female. Based on the different classifier performances the highest one is SVM acquired 98.48% also the nearest second is XG-boost.

The purpose behind this survey is to know what is the perspective of teachers and students on online education and to predict the future of online education in Bangladesh by collaborating with their perspective. The participants are students and teachers from the school, college and university of Bangladesh. The data for this research has been collected through a google form where there are two separate sections of the question set both for teachers and students. There are already some existing works on this topic but they only focus on the perspective of the students on online education and they only show what students feel about online learning. They don't give us the data what teachers have to say about online learning and they don't also mention the steps we should take to make online learning system suitable both for teachers and students. As we know to have a proper education environment the cooperation between both students and teachers should be satisfactory. So, by our survey we tried to cover the point of view of both students and teachers and by comparing their viewpoint we will give the decision if this online education will be fruitful and if it's not then what steps we should take to make it fruitful. This research can help both students and teachers to know each other's points of view so they can participate in online learning sessions by cooperating with each other and also can help the education sectors to know what they should do to support all the students and teachers with online education.

In the recent past, a growing number of users have tolerated offensive behaviors or have witnessed objectionable activities through virtual platforms. Defamatory and abusive words are mostly used in the comment section on social networking sites that are more detrimental for children and teenagers. This type of offensive word can shape their thoughts in the wrong manners and influence their mental health. Our work targets to filter and replace the negative impression of defamatory words on children that are used on Facebook with the help of sequence to sequence RNNs along with LSTM. For doing this experiment, we have created a dataset by the comments and posts from Facebook. And also, done data pre-preprocessing, counting of vocabulary, counting of missing words, embedding words, finding the missing word, and so on. Our main focus was identifying the offensive word and filtering that word by the abstract text summarizer model and decreasing the training loss. After applying the model on the dataset, practically we have reached our target to decrease the training loss to 0.007 and are capable of making a filtered text from given input text.

In this era, many business organizations and financial institutions are mostly conducted based on their large volumes of data. To analyze large volumes of data there are many techniques. In the big data, data mining process, Machine Learning is an arising strategy to investigate the huge volumes of information and is utilized to settle on basic choices for the business and financial association. It is also used in many sectors like medical, sentiment analysis for social media comments, banking sector and so on where it first learns the data and then performs the predictive analysis. In today's world, technology is progressing and with the progression of technologies many organizations are trying to adapt their business model with many technologies. For banking and financial sectors credit risk is a great threat and to predict the credit worthiness of a customer there are many techniques that exist. In this research, we have explored the dataset of bank which contains the information of credit defaulter client and applied some supervised Machine Learning algorithms to predict the credit ability of a customer and the clients who covered the highest chance of short-dated credit recovery. Also, we have performed feature scaling techniques so that many machine learning models can behave much better and perform well. We analyzed our dataset and dropped those unnecessary features that do not affect our model performance. Then we applied the final necessary features to those Machine Learning algorithms, calculated the accuracy and compared the accuracy with other classifier algorithms. Compared with other classifiers, we have observed that Random Forest performs better and it gives 89% accuracy to the prediction of credit recovery.

Classification techniques are used to classify the given data input to their desire class or target. By using classification techniques in machine learning the data can be given their classes. By keeping that in mind we focused to our Bengali word which have Bengali prefix. There are 21 prefix in Bengali grammar which have a great impact on Bengali grammar. All Bengali people are familiar with all the Bengali prefix. To classify those words by their prefix we used most common machine learning supervised algorithms. There are more than 500 data in the data set where all the data is arranged by the Bengali prefix. Classes are given by their desired prefix. After the preprocessing of data machine learning supervised algorisms are applied. Most common supervised algorithms like KNN, DT, NB, RF, LR, XGB classifier, and SVC classifier for getting more satisfying outcome are applied. There are a lot of detailed information which are described specific segment of this paper. Random forest and support vector classifier is outreach to the prominent accuracy among more than 90%.