Percent Accuracy Research Articles

Development of multistage crop yield estimation model using machine learning and deep learning techniques.

In this research paper, machine learning techniques were applied to a multivariate meteorological time series data for estimating the wheat yield of five districts of Punjab. Wheat yield data and weather parameters over 34years were collected from the study area and the model was developed using stepwise multi-linear regression (SMLR), artificial neural network (ANN), support vector regression (SVR), random forest (RF) and deep neural network (DNN) techniques. Wheat yield estimation was done at the tillering, flowering, and grain-filling stage of the crop by considering weather variables from 46 to 4th, 46 to 8th, and 46 to 11th standard meteorological week. Weighted and unweighted Meteorological variables and yield data were used to train, test, and validate the models in R software. The evaluation results showed a consistent and promising performance of RF, SVR, and DNN models for all five districts with an overall MAPE and nRMSE value of less than 6% during validation at all three growth stages. These models exhibited outstanding performance during validation for the Faridkot, Ferozpur, and Gurdaspur districts. Based on accuracy parameters MAPE, RMSE, nRMSE, and percentage deviation, the RF model was found better followed by SVR and DNN models and, hence can be used for district-level wheat crop yield estimation at different crop growth stages.

Predicting Surface Roughness and Grinding Forces in UNS S34700 Steel Grinding: A Machine Learning and Genetic Algorithm Approach to Coolant Effects

In today’s tech world of digitalization, engineers are leveraging tools such as artificial intelligence for analyzing data in order to enhance their capability in evaluating product quality effectively. This research study adds value by applying algorithms and various machine learning techniques—such as support vector regression, Gaussian process regression, and artificial neural networks—on a dataset related to the grinding process of UNS S34700 steel. What sets this study apart is its consideration of factors like three types of grinding wheels, four distinct cooling solutions, and seven varied depths of cut. These parameters are assessed for their impact on surface roughness and grinding forces, resulting in the conversion of information into insights. A relational equation with 25 coefficients is developed, using optimized algorithms to predict surface roughness with an 85 percent accuracy and grinding forces with a 90 percent accuracy rate. Learning from machine models like the Gaussian process regression exhibited stability, with an R2 value of 0.98 and a mean accuracy of 93 percent. Artificial neural networks achieved an R2 value of 0.96, and an accuracy rate of 90 percent. These findings suggest that machine learning techniques are versatile and precise when dealing with datasets. They align well with digitalization and predictive trends. In conclusion; machine learning provides flexibility and superior accuracy for predicting data trends compared to the formulaic approach, which is contained to existing datasets only. The versatility of machine learning highlights its significance in engineering practices for making data-informed decisions.

A fact based analysis of decision trees for improving reliability in cloud computing

The popularity of cloud computing (CC) has increased significantly in recent years due to its cost-effectiveness and simplified resource allocation. Owing to the exponential rise of cloud computing in the past decade, many corporations and businesses have moved to the cloud to ensure accessibility, scalability, and transparency. The proposed research involves comparing the accuracy and fault prediction of five machine learning algorithms: AdaBoostM1, Bagging, Decision Tree (J48), Deep Learning (Dl4jMLP), and Naive Bayes Tree (NB Tree). The results from secondary data analysis indicate that the Central Processing Unit CPU-Mem Multi classifier has the highest accuracy percentage and the least amount of fault prediction. This holds for the Decision Tree (J48) classifier with an accuracy rate of 89.71% for 80/20, 90.28% for 70/30, and 92.82% for 10-fold cross-validation. Additionally, the Hard Disk Drive HDD-Mono classifier has an accuracy rate of 90.35% for 80/20, 92.35% for 70/30, and 90.49% for 10-fold cross-validation. The AdaBoostM1 classifier was found to have the highest accuracy percentage and the least amount of fault prediction for the HDD Multi classifier with an accuracy rate of 93.63% for 80/20, 90.09% for 70/30, and 88.92% for 10-fold cross-validation. Finally, the CPU-Mem Mono classifier has an accuracy rate of 77.87% for 80/20, 77.01% for 70/30, and 77.06% for 10-fold cross-validation. Based on the primary data results, the Naive Bayes Tree (NB Tree) classifier is found to have the highest accuracy rate with less fault prediction of 97.05% for 80/20, 96.09% for 70/30, and 96.78% for 10 folds cross-validation. However, the algorithm complexity is not good, taking 1.01 seconds. On the other hand, the Decision Tree (J48) has the second-highest accuracy rate of 96.78%, 95.95%, and 96.78% for 80/20, 70/30, and 10-fold cross-validation, respectively. J48 also has less fault prediction but with a good algorithm complexity of 0.11 seconds. The difference in accuracy and less fault prediction between NB Tree and J48 is only 0.9%, but the difference in time complexity is 9 seconds. Based on the results, we have decided to make modifications to the Decision Tree (J48) algorithm. This method has been proposed as it offers the highest accuracy and less fault prediction errors, with 97.05% accuracy for the 80/20 split, 96.42% for the 70/30 split, and 97.07% for the 10-fold cross-validation.

Leveraging AI for Effective Fake News Detection and Verification

The rapid proliferation of fake news, misinformation, and disinformation poses a tremendous threat to the integrity of news and media production. This hazard is augmented by the instantaneous capacity to disseminate fallacious content globally with simply a touch of a button. As a result of this comprehension, this study aims to enhance the efficiency of fact-checking through the application of artificial intelligence (AI) by developing AI models to accurately classify trustworthy news, which improves the detection and containment of fake news. Additionally, the study seeks to raise awareness regarding the importance of news verification, which is achieved by ensuring information validity via AI and providing reports that evaluate the effectiveness of these novel techniques. We experimented with several machine learning models and determined the Support Vector Machine (SVM) was the best performing, which classified news articles with 98.60 percent accuracy by analyzing the first one thousand characters in news articles.

Phishing detection using clustering and machine learning

<span lang="EN-US">Phishing is a prevalent and evolving cyber threat that continues to exploit human vulnerability to deceive individuals and organizations into revealing sensitive information. Phishing attacks encompass a range of tactics, from deceptive emails and fraudulent websites to social engineering techniques. Traditional methods of detection, such as signature-based approaches and rule-based filtering, have proven to be limited in their effectiveness, as attackers frequently adapt and create new, previously unseen phishing campaigns. Consequently, there is a growing need for more sophisticated and adaptable detection methods. In recent years, Machine Learning (ML) and Artificial Intelligence (AI) have played a significant role in enhancing phishing detection. These technologies leverage large datasets to train models capable of recognizing subtle patterns and anomalies in both email content and website behavior. This research proposes a hybrid algorithm to detect phishing attack based on clustering and classification ML methods (CMLM): Deep Learning (DL) and Decision Tree (DT). Simulation results show that the proposed technique achieves high percentage of accuracy in detecting phishing</span>

Development & validation of a simple, sensitive, robust, reproducible & economical spectrofluorimetric method for estimation of taurine in novel chewable gel

BackgroundTaurine, known as 2-aminoethanesulphonic acid, is a sulfur-containing β-amino acid not incorporated into proteins. However, it takes part in biochemical reactions of significant importance, such as conjugation with bile acids to form bile salts essential for fat absorption, cell membrane stabilization, antioxidation, detoxification, osmoregulation, neuromodulation, and brain and retinal development. ObjectiveTo increase its dissolution and absorption rate for early onset of action, we developed a medicated gel formulation of taurine. Chewable gels are suitable dosage forms for pediatric, geriatric, and dysphagic patients. The current research aims to develop a simple, sensitive, and reproducible spectrofluorometric method for determining taurine in chewable gel. MethodA spectrofluorimetric method was developed, and the determination of taurine was based on Hantzsch condensation synthesis, and the fluorescence intensity was measured at emission wavelength (λem) of 474 nm after excitation (λex) at 416 nm. The spectrofluorimetry method for taurine was validated according to the Q2B(R1) ICH guideline for method validation and validated for linearity, LOD, LOQ, interday and intraday precision, accuracy, and robustness. ResultsThe calibration curve was linear in a 1.5–50 µg/ml concentration range. The correlation coefficient (r2) was found to be 0.994. The % RSD values of intra-day precision lie between 0.11 and 0.09; for inter-day precision, it was found to be between 0.09 and 0.11; in both cases, it is less than 2 %. The percentage accuracy study was between 91.81 and 99.93 with a % RSD value of 0.078, less than 2 %. LOD & LOQ values have been mathematically calculated to 4.03 µg/ml and 12.23 µg/ml, respectively. The method was precise, accurate, and robust. ConclusionAll validation parameters were found to be within their acceptable limits. Thus, the developed spectrofluorimetric method of taurine is economical, fast, and robust. Further, this method was utilized to estimate taurine content uniformity in chewable gel for in-vitro release studies.

Computational Analysis of Tsunami Wave Run-Up by Implementation of TIMPULSE-SIM Model with Japan and Indonesian Seismic Tsunami

The alternative newly developed TIMPULSE-SIM model is used for forecasting all three phases of seismic/earthquake-induced tsunami waves. The main objective of this modeling is to predict the earliest arrival time of tsunami with less computation time. This paper analyses the third phase or the run-up phase of tsunami. The assessment of tsunami wave run-up minimize the risk in coastal community due to the tsunami impact. The paper introduces a closed set of algebraic expressions for modeling the run-up phase and the reliability of the model is analyzed by implementing and testing of this model to the two major historical seismic tsunami are, 2004 Indonesian Subduction zone tsunami in Indian ocean region, 2011 Great east Japan tsunami in Pacific ocean region. From this study, this proposed model should be a good alternative to the existing model for applying to the real time tsunami event because it includes the nonlinearity and frequency dispersion of wave, and applied to both near and far field tsunamis, efficient to apply for a long duration, mainly the computation time is achieved in O(minutes). More than 90 percentage of accuracy achieved in this model by validating the simulated results with the historical and other existing model that emphasize the model’s reliability.

Effect of Instruction and Target Position on Penalty Kicking Performance in Soccer

The purpose of this study was to investigate the effect of instructions that prioritize either speed or accuracy in experienced senior football players when taking penalty kicks at five different targets in a goal. Sixteen male experienced senior football players performed in total 80 penalty kicks with instructions that prioritized either precision or speed at five different targets in the goal. Ball velocity and hit accuracy were evaluated between the two instructions and the five targets. The main findings showed that aiming for velocity resulted in higher ball velocity and lower kicking accuracy than aiming for accuracy. However, kicking accuracy was only lower when kicking to the bottom corner targets. Furthermore, when shooting high at the middle, the percentage of balls on target was higher when aiming for speed rather than accuracy. Based upon the findings of the study, it is suggested that a player should try to kick as fast as possible and aim for the middle of the goal, as ball velocity is faster than when aiming for accuracy and hit percentage in the goal is also at its highest, thereby increasing the chances of outdoing the goalkeeper, who most often goes for one of the sides.

APPLICATION OF THE ITEM RESPONSE THEORY IN IMAGE PROCESSING

The Item Response Theory (IRT) was applied to image processing where the item's difficulty parameter was used to reconstruct a region of interest in an image contaminated with noise. Shannon's entropy was used for the dichotomization of the image pixels. To demonstrate the efficiency of the proposed method, it was used on simulated data that can be related to functional magnetic resonance imaging (fMRI). The results on simulated data showed that IRT achieved a high percentage of accuracy in identifying the region of interest in the image and that, on average, the estimation converges to the true region of interest, thus proving to be a promising method for the analysis of such data.

BIOANALYTICAL METHOD DEVELOPMENT AND VALIDATION FOR THE DETERMINATION OF ASCIMINIB ANTICANCER DRUG IN BIOLOGICAL MATRICES BY LC–ESI-MS/MS

Objective: A unique liquid chromatography-mass spectrometry technique is essential for determining the concentration of asciminib in biological matrices, and its development is of the utmost importance. Methods: The samples that were processed were separated using a Reversed Phase-Phenomenex (100 mm x 4.6 mm, 5 µm) C18 analytical column. The column was equipped with an isocratic moveable phase that consisted of 0.1% (v/v) HCOOH and acetonitrile at a ratio of 18:82% (v/v). The flow rate of the phase was 0.70 ml/min. For asciminib, the multiple reaction monitoring mode was used at m/z 450.23/257.3, while for canagliflozin, it was used at m/z 445.13/267.31. Results: With a correlation coefficient of 0.9998, the method was linear for asciminib throughout the concentration range of 1.0-2100.00 ng/ml. Each day's accuracy percentage relative standard deviation was within 5.74%. For analytes at the low-quality control level, the mean matrix factors ranged from 96.34 to 104.85% with a % Coefficient of Variance (CV) of 4.21; at the high-quality control level, the range was from 94.62 to 103.88% with a %CV of 3.67. Conclusion: The method that has been developed has the potential to be used to examine the pharmacokinetics and toxicokinetics of asciminib in various biological samples for both forensic and clinical purposes.

Accuracy of Prospective Assessments of 4 Large Language Model Chatbot Responses to Patient Questions About Emergency Care: Experimental Comparative Study.

Recent surveys indicate that 48% of consumers actively use generative artificial intelligence (AI) for health-related inquiries. Despite widespread adoption and the potential to improve health care access, scant research examines the performance of AI chatbot responses regarding emergency care advice. We assessed the quality of AI chatbot responses to common emergency care questions. We sought to determine qualitative differences in responses from 4 free-access AI chatbots, for 10 different serious and benign emergency conditions. We created 10 emergency care questions that we fed into the free-access versions of ChatGPT 3.5 (OpenAI), Google Bard, Bing AI Chat (Microsoft), and Claude AI (Anthropic) on November 26, 2023. Each response was graded by 5 board-certified emergency medicine (EM) faculty for 8 domains of percentage accuracy, presence of dangerous information, factual accuracy, clarity, completeness, understandability, source reliability, and source relevancy. We determined the correct, complete response to the 10 questions from reputable and scholarly emergency medical references. These were compiled by an EM resident physician. For the readability of the chatbot responses, we used the Flesch-Kincaid Grade Level of each response from readability statistics embedded in Microsoft Word. Differences between chatbots were determined by the chi-square test. Each of the 4 chatbots' responses to the 10 clinical questions were scored across 8 domains by 5 EM faculty, for 400 assessments for each chatbot. Together, the 4 chatbots had the best performance in clarity and understandability (both 85%), intermediate performance in accuracy and completeness (both 50%), and poor performance (10%) for source relevance and reliability (mostly unreported). Chatbots contained dangerous information in 5% to 35% of responses, with no statistical difference between chatbots on this metric (P=.24). ChatGPT, Google Bard, and Claud AI had similar performances across 6 out of 8 domains. Only Bing AI performed better with more identified or relevant sources (40%; the others had 0%-10%). Flesch-Kincaid Reading level was 7.7-8.9 grade for all chatbots, except ChatGPT at 10.8, which were all too advanced for average emergency patients. Responses included both dangerous (eg, starting cardiopulmonary resuscitation with no pulse check) and generally inappropriate advice (eg, loosening the collar to improve breathing without evidence of airway compromise). AI chatbots, though ubiquitous, have significant deficiencies in EM patient advice, despite relatively consistent performance. Information for when to seek urgent or emergent care is frequently incomplete and inaccurate, and patients may be unaware of misinformation. Sources are not generally provided. Patients who use AI to guide health care decisions assume potential risks. AI chatbots for health should be subject to further research, refinement, and regulation. We strongly recommend proper medical consultation to prevent potential adverse outcomes.

Measurements of F1‐ Region Ionosphere State Variables at Arecibo Through Quasi Height‐Independent Exhaustive Fittings of the Incoherent Scatter Ion‐Line Spectra

AbstractWe discuss an exhaustive search approach to fit the incoherent scatter spectrum (ISS) in the F1‐region for molecular ion fraction (fm), ion temperature (Ti), and electron temperature (Te). The commonly used “full profile” approach for F1‐region measurements parameterizes the molecular ion fraction as a function of altitude and fits all the related heights for the state variables. In our approach, we fit the ISS at each height for fm, Ti, Te, and ion velocity (Vi) independently. Our exhaustive search method finds all the major local minima at each altitude. Although a parameterized function is used to guide the algorithm in finding the best solution, the fitting parameters retain their local characteristics. Despite that fitting fm, Ti, and Te without constraints requires Doppler shift to be accurately determined and the ISS signal‐to‐noise ratio higher than the full‐profile method, simulations show that Ti, Te, and fm can be recovered within a few percent accuracy with a moderate signal‐to‐noise ratio. We apply the exhaustive search approach to the Arecibo high‐resolution incoherent scatter radar data taken on 13 September 2014. The derived ion and electron temperatures are sensitive enough to reveal thermosphere gravity waves commonly seen in the electron density previously. Our method is more robust than previous height‐independent fitting methods. Comparison with another Arecibo program indicates our results are likely more accurate. Simultaneous high‐resolution measurements of Ti, Te, fm, Vi, and electron concentration (Ne) in our approach open new opportunities for synergistic studies of the F1‐region dynamics and chemistry.

Experimental performance verification of an intelligent detection and assessment scheme for disturbances and imbalances of three-phase synchronous machine output using coherence estimators

In this article, power quality disturbances, such as voltage and current unbalances, series and shunt faults, are monitored in three-phase synchronous machines. Because the imbalances affect machine efficiency and service life, it is essential to figure out whether the three-phase voltages/currents are balanced or unbalanced. An integrated detection algorithm is proposed to detect the unbalanced voltages/currents and various abnormal conditions precisely and quickly, which is based on the coherence estimator. Most existing detection methods require more than one cycle time to determine the contingency conditions, while the proposed technique detects these situations in a fast and discrete protection system. The suggested scheme acquires instantaneous measurements of the three-phase voltages and currents taken at the synchronous machine terminals in order to calculate fifteen coherence coefficients that are utilized to detect and assess any unbalance or trouble accurately and swiftly. A new proposal for imbalance and disturbance indicators based on the coherence estimators is developed in this study. Multiple test cases have been conducted to validate the proposed algorithm capabilities using a real power system, which includes a motor-generator set, a three-phase load and measurement transformers. The experimental results have been revealed that the relay reliability and accuracy percentages have been found to be 98.28% and 98.52%, respectively. The quantitative findings of the imbalance and disturbance assessment are recorded.

Prescribing Blood Flow Restricted Exercise: Limb Composition Influences the Pressure Required to Create Arterial Occlusion.

As blood flow restriction gains popularity across different populations (eg,young and older adults) and settings (eg,clinical and sports rehabilitation), the accuracy of blood flow restricted percentage becomes crucial. We aimed to compare manually measured arterial occlusion pressure (AOP) among young adults to understand whether lower limb composition affects the pressure required to achieve AOP. The results will shed light on the adequacy of published calculations used to estimate AOP in practical and research settings. An observational cross-sectional study design was implemented to examine the relationship between lower limb composition, lower limb circumference, and measured AOP. Twenty-two participants (12 males, 26 [4]y, 1.74 [0.07]m, 73.2 [12.5]kg) underwent a whole-body Dual-energy X-ray Absorptiometry scan before AOP (in millimeters of mercury) and lower limb circumference (in centimeters) were determined. In a supine position, a 10-cm wide cuff was manually inflated on the dominant leg to the point where a pulse could no longer be detected by a Doppler ultrasound of the posterior tibial artery to determine AOP. Lower limb composition (fat, muscle, and bone mass [in grams]) was obtained from the Dual-energy X-ray Absorptiometry scan. Lower limb muscle mass had a moderate negative relationship with AOP (r2 = .433, β = -0.004) and a moderate positive relationship with lower limb circumference (r2 = .497, β = 0.001). Lower limb circumference had the weakest relationship with AOP (r2 = .316, β = 0.050) of all measures. The reported relationships between lower limb muscle mass, lower limb circumference, and AOP suggest that as muscle mass increases, lower limb circumference also increases, yet AOP decreases. This implies that limb circumference should not be used as the primary measure for calculating AOP within the sampled population. We recommend individually measuring AOP when implementing blood flow restriction in all exercise modalities.

APPLICATION OF THE QUEST AND CHAID METHODS IN CLASSIFYING STUDENT GRADUATION

Graduation is the final result of the learning process during the course. Student graduation time is affected by many factors. Whether or not the time of student graduation is appropriate is an important thing that must be considered. Graduating well and on time is one measure of success in the learning process. This research aims to build a student graduation classification model by applying the QUEST (Quick, Unbiased, and Efficient, Statistical Tree) and CHAID (Chi-squared Automatic Interaction Detection) methods, examining the factors that affect student graduation, and comparing the classification results of the two methods. Both methods produce output in the form of tree diagrams, making it easier to interpret. Based on the classification tree formed from the two methods, four final nodes of the classification tree were generated, and three categories were grouped. Factors that affect student graduation include age and IPK. The classification results show that the percentage of classification accuracy for student graduation with QUEST and CHAID methods is 76.1%.

Investigating enhanced electrical conductivity for antenna applications through dual metallization on 3D printed SLA substrates

The advancement of 3D printing (additive manufacturing) has gained interest in variety of applications, especially for antenna fabrication. The demand for cheap, reliable, and high-performance antenna fabrication methods is essential in order to cope with the demand of wireless communications industry. In this work, the focus was emphasized on enhancing the electrical conductivity of the 3D printed substrates fabricated by stereolithography (SLA) 3D printer. The 3D printed substrate went through a dual metallization approach, involving sputtering and electrodeposition techniques for the fabrication of conductive metal layers. The parameters for the sputtering were fixed for all the substrate while current density during electrodeposition process was varied at 25 %, 50 %, 75 % and 100 % from recommended value of current. The results showed that the current density during the electrodeposition determined the conductivity performance of the metal layers and established a significant correlation with the surface morphological. Three different frequencies comprised of 2.4 GHz, 3.5 GHz and 5.0 GHz were selected to simulate and fabricate a microstrip patch antenna using the optimized current density which was valued optimal at 75 % (52.5 mA). The percentage of accuracy between simulated and measured frequencies of antenna, portrayed that the measured values were slightly higher than the simulated approximately about 5.14 to 6.67 %. Therefore, the integration of additive manufacturing or 3D printing techniques for antenna could address the critical necessity for fast and economical solution in wireless systems.

Repeated time-series cross-validation: A new method to improved COVID-19 forecast accuracy in Malaysia

Forecasting COVID-19 cases is challenging, and inaccurate forecast values will lead to poor decision-making by the authorities. Conversely, accurate forecasts aid Malaysian government authorities and agencies (National Security Council, Ministry of Health, Ministry of Finance, Ministry of Education, and Ministry of International Trade and Industry) and financial institutions in formulating action plans, regulations, and legal acts to control COVID-19 spread in the country. Therefore, this study proposes Repeated Time-Series Cross-Validation, a new data-splitting strategy to identify the best forecasting model that is capable of producing the lowest error measures value and a high percentage of forecast accuracy for COVID-19 prediction in Malaysia. Some of the highlights of the proposed method are:•A total of 21 models, five data partitioning sets, and four error measures to improve the forecast accuracy of daily COVID-19 cases in Malaysia.•The best model selected produces the lowest error measure value for the Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Mean Absolute Scaled Error (MASE).•The average 8-day forecast accuracy is 90.2 %. The lowest and highest forecast accuracy was 83.7 % and 98.7 %.

Morphometric Assessment of the Piriform Aperture and Its Clinical and Forensic Applications

Abstract Background: The piriform aperture (PA) is a midface structure that forms the anterior skeletal boundary of the nose. It is highly variant due to its physiological adaptations to the climate. Its variations are useful in forensic identification and reconstructive surgeries of the face. This study aimed at determining the PA’s dimensions and their accuracy in sex determination. Materials and Methods: The PA’s dimensions were retrospectively examined using 336 (199 males and 137 females) adult cranial computed tomography images in the database of the radiology unit of a university teaching hospital in Delta State, Nigeria, following institutional authorization. The aperture’s index was calculated as a ratio of height to width. The data were analyzed using Statistical Package for the Social Sciences Version 23. The sexual dimorphism in the parameters was assessed using an independent t-test. Association among the continuous variables was analyzed by Pearson’s correlation test. P = 0.05 was considered statistically significant. The percentage accuracy for correct sex prediction was assessed using the discriminant function analysis. Results: A significant relationship between dimensions with sex was observed (P < 0.05). The width of the aperture was the best sex-discriminating parameter (70.2%). The overall accuracy for sex discrimination using the aperture’s dimensions was 75.0%. Conclusion: This study provides the standard ranges for the PA’s width and height, valuable for surgical planning. These dimensions were sexually dimorphic and demonstrated an acceptable overall accuracy of correct sex allocation (75%). Consequently, this aperture may be utilized as a supplementary tool in conjunction with other methods for sex determination within studied population.

Deep Convolutional Transformer Network for Stock Movement Prediction

The prediction and modeling of stock price movements have been shown to possess considerable economic significance within the finance sector. Recently, a range of artificial intelligence methodologies, encompassing both traditional machine learning and deep learning approaches, have been introduced for the purpose of forecasting stock price fluctuations, yielding numerous successful outcomes. Nonetheless, the identification of effective features for predicting stock movements is considered a complex challenge, primarily due to the non-linear characteristics, volatility, and inherent noise present in financial data. This study introduces an innovative Deep Convolutional Transformer (DCT) model that amalgamates convolutional neural networks, Transformers, and a multi-head attention mechanism. It features an inception convolutional token embedding architecture alongside separable fully connected layers. Experiments conducted on the NASDAQ, Hang Seng Index (HSI), and Shanghai Stock Exchange Composite (SSEC) employ Mean Absolute Error (MAE), Mean Square Error (MSE), Mean Absolute Percentage Error (MAPE), accuracy, and Matthews Correlation Coefficient (MCC) as evaluation metrics. The findings reveal that the DCT model achieves the highest accuracy of 58.85% on the NASDAQ dataset with a sliding window width of 30 days. In terms of error metrics, it surpasses other models, demonstrating the lowest average prediction error across all datasets for MAE, MSE, and MAPE. Furthermore, the DCT model attains the highest MCC values across all three datasets. These results suggest a promising capability for classifying stock price trends and affirming the DCT model’s superiority in predicting closing prices.

OUCH: Oversampling and Undersampling Cannot Help Improve Accuracy in Our Bayesian Classifiers That Predict Preeclampsia

Unbalanced data can have an impact on the machine learning (ML) algorithms that build predictive models. This manuscript studies the influence of oversampling and undersampling strategies on the learning of the Bayesian classification models that predict the risk of suffering preeclampsia. Given the properties of our dataset, only the oversampling and undersampling methods that operate with numerical and categorical attributes will be taken into consideration. In particular, synthetic minority oversampling techniques for nominal and continuous data (SMOTE-NC), SMOTE—Encoded Nominal and Continuous (SMOTE-ENC), random oversampling examples (ROSE), random undersampling examples (UNDER), and random oversampling techniques (OVER) are considered. According to the results, when balancing the class in the training dataset, the accuracy percentages do not improve. However, in the test dataset, both positive and negative cases of preeclampsia were accurately classified by the models, which were built on a balanced training dataset. In contrast, models built on the imbalanced training dataset were not good at detecting positive cases of preeclampsia. We can conclude that while imbalanced training datasets can be addressed by using oversampling and undersampling techniques before building prediction models, an improvement in model accuracy is not always guaranteed. Despite this, the sensitivity and specificity percentages improve in binary classification problems in most cases, such as the one we are dealing with in this manuscript.