Supervised Machine Learning Methods Research Articles

Quantitative Precipitation Estimates Using Machine Learning Approaches with Operational Dual-Polarization Radar Data

Traditional radar-based rainfall estimation is typically done by known functional relationships between the rainfall intensity (R) and radar measurables, such as R–Zh, R–(Zh, ZDR), etc. One of the biggest advantages of machine learning algorithms is the applicability to a non-linear relationship between a dependent variable and independent variables without any predefined relationships. We explored the potential use of two supervised machine learning methods (regression tree and random forest) in rainfall estimation using dual-polarization radar variables. The regression tree does not require normalization and scaling of data; however, this method is quite unstable since each split depends on the parent split. Since the random forest is an ensemble method of regression trees, it has less variability in prediction compared with regression trees, but consumes more computer resources. We considered several different configurations for machine learning algorithms with different sets of dependent and independent variables. The random forest model was appropriately tuned. In the test of variable importance, the specific differential phase (differential reflectivity) was the most important variable to predict the rainfall rate (residual that is the difference between the true rainfall rate and the one estimated from the R–Z relationship). The models were evaluated by 10-fold cross-validation. The best model was the random forest model using a residual with the non-classified training set. The results indicated that the machine learning algorithms outperformed the traditional R–Z relationship. Then, we applied the best machine learning model to an S-band dual-polarization radar (Mt. Myeonbong) and validated the result with ground rain gauges. The results of the application to radar data showed that the estimates of the residuals had spatial variability. The stratiform and weak rain areas had positive residuals while convective areas had negative residuals, indicating that the spatial error structure driven by the R–Z relationship was well captured by the model. The rainfall rates of all pixels over the study area were adjusted with the estimated residuals. The rainfall rates adjusted by residual showed excellent agreement with the rain gauge, especially at high rainfall rates.

A Machine Learning Approach to Predicting New-onset Depression in a Military Population.

Objective:Depression is one of the most common mental disorders in the United States in both civilian and military populations, but few prospective studies assess a wide range of predictors across multiple domains for new-onset (incident) depression in adulthood. Supervised machine learning methods can identify predictors of incident depression out of many different candidate variables, without some of the assumptions and constraints that underlie traditional regression analyses. The objectives of this study were to identify predictors of incident depression across 5 years of follow-up using machine learning, and to assess prediction accuracy of the algorithms.Methods:Data were from a cohort of Army National Guard members free of history of depression at baseline (n = 1951 men and 298 women), interviewed once per year for probable depression. Classification trees and random forests were constructed and cross-validated, using 84 candidate predictors from the baseline interviews.Results:Stressors and traumas such as emotional mistreatment and adverse childhood experiences, demographics such as being a parent or student, and military characteristics including paygrade and deployment location were predictive of probable depression. Cross-validated random forest algorithms were moderately accurate (68% for women and 73% for men).Conclusions:Events and characteristics throughout the life course, both in and outside of deployment, predict incident depression in adulthood among military personnel. Although replication studies are needed, these results may help inform potential intervention targets to reduce depression incidence among military personnel. Future research should further refine and explore interactions between identified variables.

Machine-learning-based EDFA gain estimation [Invited

Optical transmission systems with high spectral efficiency require accurate quality of transmission estimation for optical channel provisioning. However, the wavelength-dependent gain effects of erbium-doped fiber amplifiers (EDFAs) complicate precise optical channel power prediction and low-margin operation. In this work, we examine supervised machine learning methods using multiple artificial neural networks (ANNs) to build models for gain spectra prediction of optical transmission line EDFAs under different operating conditions. Channel-loading configurations and channel input power spectra are used as an a posteriori knowledge data feature for model training. In a hybrid learning approach, estimated gain spectra calculated by an analytical model are added as an a priori input data feature to further improve the EDFA ANN model performance in terms of prediction accuracy, training time, and quantity of training data. Using these methods, the root mean square error and maximum absolute error of the predicted channel output power can be as low as 0.144 dB and 1.6 dB, respectively.

Relationship Between Wellness Index and Internal Training Load in Soccer: Application of a Machine Learning Model.

To investigate the relationship between the training load (TL = rate of perceived exertion × training time) and wellness index (WI) in soccer. The WI and TL data were recorded from 28 subelite players (age = 20.9 [2.4]y; height = 181.0 [5.8]cm; body mass = 72.0 [4.4]kg) throughout the 2017/2018 season. Predictive models were constructed using a supervised machine learning method that predicts the WI according to the planned TL. The validity of our predictive model was assessed by comparing the classification's accuracy with the one computed from a baseline that randomly assigns a class to an example by respecting the distribution of classes (B1). A higher TL was reported after the games and during match day (MD)-5 and MD-4, while a higher WI was recorded on the following days (MD-6, MD-4, and MD-3, respectively). A significant correlation was reported between daily TL (TLMDi) and WI measured the day after (WIMDi+1) (r = .72, P < .001). Additionally, a similar weekly pattern seems to be repeating itself throughout the season in both TL and WI. Nevertheless, the higher accuracy of ordinal regression (39% [2%]) compared with the results obtained by baseline B1 (21% [1%]) demonstrated that the machine learning approach used in this study can predict the WI according to the TL performed the day before (MD<i). The machine learning technique can be used to predict the WI based on a targeted weekly TL. Such an approach may contribute to enhancing the training-induced adaptations, maximizing the players' readiness and reducing the potential drops in performance associated with poor wellness scores.

Prediction of Pathological Upgrading at Radical Prostatectomy in Prostate Cancer Eligible for Active Surveillance: A Texture Features and Machine Learning-Based Analysis of Apparent Diffusion Coefficient Maps.

ObjectiveTo evaluate a combination of texture features and machine learning-based analysis of apparent diffusion coefficient (ADC) maps for the prediction of Grade Group (GG) upgrading in Gleason score (GS) ≤6 prostate cancer (PCa) (GG1) and GS 3 + 4 PCa (GG2).Materials and methodsFifty-nine patients who were biopsy-proven to have GG1 or GG2 and underwent MRI examination with the same MRI scanner prior to transrectal ultrasound (TRUS)-guided systemic biopsy were included. All these patients received radical prostatectomy to confirm the final GG. Patients were divided into training cohort and test cohort. 94 texture features were extracted from ADC maps for each patient. The independent sample t-test or Mann−Whitney U test was used to identify the texture features with statistically significant differences between GG upgrading group and GG non-upgrading group. Texture features of GG1 and GG2 were compared based on the final pathology of radical prostatectomy. We used the least absolute shrinkage and selection operator (LASSO) algorithm to filter features. Four supervised machine learning methods were employed. The prediction performance of each model was evaluated by area under the receiver operating characteristic curve (AUC). The statistical comparison between AUCs was performed.ResultsSix texture features were selected for the machine learning models building. These texture features were significantly different between GG upgrading group and GG non-upgrading group (P < 0.05). The six features had no significant difference between GG1 and GG2 based on the final pathology of radical prostatectomy. All machine learning methods had satisfactory predictive efficacy. The diagnostic performance of nearest neighbor algorithm (NNA) and support vector machine (SVM) was better than random forests (RF) in the training cohort. The AUC, sensitivity, and specificity of NNA were 0.872 (95% CI: 0.750−0.994), 0.967, and 0.778, respectively. The AUC, sensitivity, and specificity of SVM were 0.861 (95%CI: 0.732−0.991), 1.000, and 0.722, respectively. There had no significant difference between AUCs in the test cohort.ConclusionA combination of texture features and machine learning-based analysis of ADC maps could predict PCa GG upgrading from biopsy to radical prostatectomy non-invasively with satisfactory predictive efficacy.

Detecting virtual concept drift of regressors without ground truth values

Regression analysis is a standard supervised machine learning method used to model an outcome variable in terms of a set of predictor variables. In most real-world applications the true value of the outcome variable we want to predict is unknown outside the training data, i.e., the ground truth is unknown. Phenomena such as overfitting and concept drift make it difficult to directly observe when the estimate from a model potentially is wrong. In this paper we present an efficient framework for estimating the generalization error of regression functions, applicable to any family of regression functions when the ground truth is unknown. We present a theoretical derivation of the framework and empirically evaluate its strengths and limitations. We find that it performs robustly and is useful for detecting concept drift in datasets in several real-world domains.

Using weak supervision and deep learning to classify clinical notes for identification of current suicidal ideation

Mental health concerns, such as suicidal thoughts, are frequently documented by providers in clinical notes, as opposed to structured coded data. In this study, we evaluated weakly supervised methods for detecting “current” suicidal ideation from unstructured clinical notes in electronic health record (EHR) systems. Weakly supervised machine learning methods leverage imperfect labels for training, alleviating the burden of creating a large manually annotated dataset. After identifying a cohort of 600 patients at risk for suicidal ideation, we used a rule-based natural language processing approach (NLP) approach to label the training and validation notes (n = 17,978). Using this large corpus of clinical notes, we trained several statistical machine learning models—logistic classifier, support vector machines (SVM), Naive Bayes classifier—and one deep learning model, namely a text classification convolutional neural network (CNN), to be evaluated on a manually-reviewed test set (n = 837). The CNN model outperformed all other methods, achieving an overall accuracy of 94% and a F1-score of 0.82 on documents with “current” suicidal ideation. This algorithm correctly identified an additional 42 encounters and 9 patients indicative of suicidal ideation but missing a structured diagnosis code. When applied to a random subset of 5,000 clinical notes, the algorithm classified 0.46% (n = 23) for “current” suicidal ideation, of which 87% were truly indicative via manual review. Implementation of this approach for large-scale document screening may play an important role in point-of-care clinical information systems for targeted suicide prevention interventions and improve research on the pathways from ideation to attempt.

Selection of optimal hyper-parameter values of support vector machine for sentiment analysis tasks using nature-inspired optimization methods

Sentiment analysis and classification task is used in recommender systems to analyze movie reviews, tweets, Facebook posts, online product reviews, blogs, discussion forums, and online comments in social networks. Usually, the classification is performed using supervised machine learning methods such as support vector machine (SVM) classifier, which have many distinct parameters. The selection of the values for these parameters can greatly influence the classification accuracy and can be addressed as an optimization problem. Here we analyze the use of three heuristics, nature-inspired optimization techniques, cuckoo search optimization (CSO), ant lion optimizer (ALO), and polar bear optimization (PBO), for parameter tuning of SVM models using various kernel functions. We validate our approach for the sentiment classification task of Twitter dataset. The results are compared using classification accuracy metric and the Nemenyi test.

A Predictive Analytics Model for Students Grade Prediction by Supervised Machine Learning

Research on predictive analytics has increasingly evolved due to its impact on providing valuable and intuitive feedback that could potentially assist educators in improving student success in higher education. By leveraging predictive analytics, educators could design an effective mechanism to improve the academic results to prevent students’ dropout and assure student retention. Hence, this paper aims to presents a predictive analytics model using supervised machine learning methods that predicts the student’s final grade (FG) based on their historical academic performance of studies. The work utilized dataset gathered from 489 students of Information and Communication Technology Department at north-western Malaysia Polytechnic over the four past academic years, from 2016 to 2019. We carried out the experiments using Decision Tree (J48), Random Forest (RF), Support Vector Machines (SVM), and Logistic Regression (LR) to study the comparison performance for both classification and regression techniques in predicting students FG. The findings from the results present that J48 was the best predictive analytics model with the highest prediction accuracy rate of 99.6% that could contribute to the early detection of students’ dropout so that educators can remain the outstanding achievement in higher education.

Review on Anatomical Medical Images Classification Methods: Diagnostic Value

Deep learning (DL) based convolutional neural network (CNN) has grown rapidly and become a selected choice for medical imaging fields. The paper reviews into three categories; 1) exploring the supervised machine learning methods, 2) reviews the limitations in deep learning, 3) and finally reviews the majors deep learning techniques within a specific summarized on lesion classification-based DL in term of (application, method, type of lesion diseases classification, cons).

A probabilistic predictive model for assessing the economic reusability of load-bearing building components: Developing a Circular Economy framework

Abstract The reuse of load-bearing building components has the potential to promote the circular economy in the building sector. One recent aspect of the efforts to improve reuse rates in buildings is estimating the reusability of the structural elements. This study develops a probabilistic predictive model using advanced supervised machine learning methods to evaluate the economic reusability of the load-bearing building elements. The results of sensitivity analysis and visualization techniques used in this study reveal that the most important economic factor is the need to purchase reused elements early in a project, which could have cash flow implications. The other most important factors are the potential financial risks, the procurement process, and the labour cost. This study unveils that the relationship between variables is not linear, and none of the identified factors could alone determine if an element is reusable or not. This study concludes that the complex interdependencies of factors affecting reuse cause a high level of uncertainty about the feasibility of reusing the load-bearing building structural components from an economic aspect. Nonetheless, this paper reveals that by using the probability theory foundations and combining it with advanced supervised machine learning methods, it is possible to develop tools that could reliably estimate the economic reusability of these elements based on affecting variables. Therefore, the authors suggest utilizing the approach developed in this research to promote the circularity of materials in different subsectors of the construction industry.

Geochemistry of the Von Kármán crater floor and thickness of the non-mare ejecta over the Chang'e-4 landing area

China's Chang'e-4 mission has carried out the first ever lunar farside landing exploration on the floor of the Von Kármán crater, a geologically complex region located in the most ancient and deepest South Pole-Aitken (SPA) basin. In order to demonstrate the characteristics of materials in the landing area, we investigated the regional geochemistry and thickness of non-mare ejecta overlaying the mare basalts. Comparative analyses of FeO, TiO2 and Th concentrations suggest that the landing site surface is dominated by non-mare ejecta from nearby craters (e.g., Finsen crater) with part of basaltic materials. The ejecta thickness is estimated based on the excavation depth of dark-haloed and non-dark-haloed craters by using support-vector machine, a supervised machine learning method for classification. The results show that the ejecta thickness in the region of 40 km across the landing site varies from near zero to ~80 m with a mean value of ~41 m. The ejecta at the Chang'e-4 landing site is ~40 m thick, which is comparable to the in situ observations by the Lunar Penetrating Radar onboard the Yutu-2 rover. Our results provide valuable information for interpretation of the on-going returned data and geologic analysis of the Chang'e-4 exploration region.

Hanehalkı Tüketim Harcamalarının Mikroekonometrik Analizi: LAD-LASSO Yöntemi

This study examined how supervised machine learning methods help us select the relevant variables of a Household Budget Survey Consumption Expenditures dataset with outliers in order to achieve better performance in the predicting and forecasting of the Household Consumption Expenditures Model. To achieve this, the Household Budget Survey Consumption Expenditures dataset of Turkey for 2018 was examined using the Least Absolute Deviation (LAD), Least Absolute Shrinkage and Selection Operator (LASSO) and LAD-LASSO methods. In addition, the classical regression method and the prediction and forecasting performances of the methods were compared. According to the analyzed results,it was concluded that the LAD-LASSO machine learning method, which enables the selection of variables while obtaining robust predictors in the presence of long-tailed errors, was the most successful method in prediction performance and forecasting accuracy. Additionally, several fundamental variables such as income, saving, and household size increase the household consumption expenditures for all models. In addition to these variables, other variables including the structure of a room, the kitchen, bathroom floors, heating, air conditioning preferences, energy sources used, detached house, apartment, cottage, vineyard ownership, investment preferences, credit card usage, and internet shopping habits were selected as determinants of household consumption expendituresin the LAD-LASSO model. From the results of the study, it wasfound that machine learning algorithms can be used in the selection of the most appropriate variablesin the course of the construction of microeconometric models.

Интеллектуальная система поддержки функциональных характеристик РЛК мониторинга на основе прогнозирования отказов

Показаны особенности применения методов искусственного интеллекта в современных радиолокационных комплексах (РЛК) мониторинга для организации превентивного технического обслуживания. Приведена структурно-функциональная схема интеллектуальной системы поддержки функциональных характеристик РЛК. Представлены экспериментальные оценки точности прогнозирования излучаемой мощности РЛК по данным диагностического контроля. Purpose. This work is devoted to a description of an intelligent support system (ICP) of the functional characteristics of radar systems operating in heat-stressed modes. The purpose of the work is to simulate the operation of the failure prediction module based on the data on the operation of the critical components of the radar. Methodology. Failure prediction based on block temperature data was performed using supervised machine learning methods, including regression methods. The highest quality parameters of the forecasting model (horizon - accuracy) were achieved when training the model using the gradient boosting method, which allows organizing continuous additional training of the forecasting model during the operation of the radar by forming a training sample consisting of a data flow about the technical state. Findings. A structural and functional diagram of the ICP has been developed and a description of the operation of its components is given. The error (20%) and the horizon (20 min.) of the prediction of failures obtained during the experiments are sufficient for the staff serving the radar to take the necessary measures and prevent a critical decrease in the functional characteristics of the product. Originality/value. The introduction of ISP will allow realizing the maximum functional capabilities of the radar, taking into account the forecast of their change, as well as making an operational report to the command post of the radar on the reasons for the decrease in functional readiness according to the pop-up tips about the faulty element and the degree of its influence on the functional characteristics.

Makine Öğrenmesi Algoritmalarını Kullanarak Rüzgar Enerjisi Üretimi Tahmini

Renewable energy becomes progressively popular in the world because renewable resources such as solar, geothermal, wind energy are clean, inexhaustible and come from natural sources. Wind energy is one of the most significant resources of renewable energy and it plays a key role in the generation of electricity. Thus, accurate wind power estimation is crucial to deal with the challenges to balance energy trading, planning, scheduling decisions and strategies of wind power generation. This study proposes a prediction model to solve a real-life problem in the renewable energy sector by accurately estimating the amount of wind energy production per hour in the next 24 hours by applying machine learning (ML) techniques using historical wind power generation data and weather forecasting reports. In the proposed approach, first, an unsupervised ML method (i.e., the K-Means clustering algorithm) is applied to group data into meaningful clusters; then, these clusters are accepted as new feature values and added to the dataset to enlarge it; finally, a supervised ML method (i.e., regression) is performed for prediction. This study compares nine supervised learning algorithms: K-Nearest Neighbors, Support Vector Regression, Random Forest, Extra Trees, Gradient Boosting, Ridge Regression, Least Absolute Shrinkage and Selection Operator, Decision Tree, and Convolutional Neural Network. The aim of this study is to investigate the success of different ML algorithms on real-world data of wind turbines and propose a methodology to benchmark various machine learning algorithms to choose the most accurate final model for wind power generation prediction.

Computational Barthel Index: an automated tool for assessing and predicting activities of daily living among nursing home patients

BackgroundAssessment of functional ability, including activities of daily living (ADLs), is a manual process completed by skilled health professionals. In the presented research, an automated decision support tool, the Computational Barthel Index Tool (CBIT), was constructed that can automatically assess and predict probabilities of current and future ADLs based on patients’ medical history.MethodsThe data used to construct the tool include the demographic information, inpatient and outpatient diagnosis codes, and reported disabilities of 181,213 residents of the Department of Veterans Affairs’ (VA) Community Living Centers. Supervised machine learning methods were applied to construct the CBIT. Temporal information about times from the first and the most recent occurrence of diagnoses was encoded. Ten-fold cross-validation was used to tune hyperparameters, and independent test sets were used to evaluate models using AUC, accuracy, recall and precision. Random forest achieved the best model quality. Models were calibrated using isotonic regression.ResultsThe unabridged version of CBIT uses 578 patient characteristics and achieved average AUC of 0.94 (0.93–0.95), accuracy of 0.90 (0.89–0.91), precision of 0.91 (0.89–0.92), and recall of 0.90 (0.84–0.95) when re-evaluating patients. CBIT is also capable of predicting ADLs up to one year ahead, with accuracy decreasing over time, giving average AUC of 0.77 (0.73–0.79), accuracy of 0.73 (0.69–0.80), precision of 0.74 (0.66–0.81), and recall of 0.69 (0.34–0.96). A simplified version of CBIT with 50 top patient characteristics reached performance that does not significantly differ from full CBIT.ConclusionDischarge planners, disability application reviewers and clinicians evaluating comparative effectiveness of treatments can use CBIT to assess and predict information on functional status of patients.

Evidence of vascular endothelial dysfunction in Wooden Breast disorder in chickens: Insights through gene expression analysis, ultra-structural evaluation and supervised machine learning methods.

Several gene expression studies have been previously conducted to characterize molecular basis of Wooden Breast myopathy in commercial broiler chickens. These studies have generally used a limited sample size and relied on a binary disease outcome (unaffected or affected by Wooden Breast), which are appropriate for an initial investigation. However, to identify biomarkers of disease severity and development, it is necessary to use a large number of samples with a varying degree of disease severity. Therefore, in this study, we assayed a relatively large number of samples (n = 96) harvested from the pectoralis major muscle of unaffected (U), partially affected (P) and markedly affected (A) chickens. Gene expression analysis was conducted using the nCounter MAX Analysis System and data were analyzed using four different supervised machine-learning methods, including support vector machines (SVM), random forests (RF), elastic net logistic regression (ENET) and Lasso logistic regression (LASSO). The SVM method achieved the highest prediction accuracy for both three-class (U, P and A) and two-class (U and P+A) classifications with 94% prediction accuracy for two-class classification and 85% for three-class classification. The results also identified biomarkers of Wooden Breast severity and development. Additionally, gene expression analysis and ultrastructural evaluations provided evidence of vascular endothelial cell dysfunction in the early pathogenesis of Wooden Breast.

Supervised machine learning methods in psychology: A practical introduction with annotated R code

AbstractMachine learning methods for prediction and pattern detection are increasingly prevalent in psychological research. We provide an introductory overview of machine learning, its applications, and describe how to implement models for research. We review fundamental concepts of machine learning, such as prediction accuracy and out‐of‐sample evaluation, and summarize standard prediction algorithms including linear regressions, ridge regressions, decision trees, and random forests (plus additional algorithms in the supplementary materials). We demonstrate each method with examples and annotated R code, and discuss best practices for determining sample sizes; comparing model performances; tuning prediction models; preregistering prediction models; and reporting results. Finally, we discuss the value of machine learning methods in maintaining psychology’s status as a predictive science.

ПРИМЕНЕНИЕ АНСАМБЛЕВЫХ МЕТОДОВ МАШИННОГО ОБУЧЕНИЯ ДЛЯ ПРОГНОЗИРОВАНИЯ ТЕХНИЧЕСКОГО СОСТОЯНИЯ ОБЪЕКТА

To ensure the reliable functioning of a technical object, it is necessary to predict its state for the upcoming time interval. Let the technical state of the object be characterized at a certain point in time by a set of parameters established by the technical documentation for the object. It is assumed that for certain values of these parameters, the object may be in a good or faulty state. It is required by the values of these parameters to estimate the state of the object in the upcoming time interval. Supervised machine learning methods can be applied to solve this problem. However, to obtain good results in predicting the state of an object, it is necessary to choose the correct training model. One of the disadvantages of machine learning models is high bias and too much scatter. In this paper, to reduce the scatter of the model, it is proposed to use ensemble machine learning methods, namely, the bagging procedure. The main idea of the ensemble of methods is that with the right combination of weak models, more accurate and robust models can be obtained. The purpose of bagging is to create an ensemble model that is more reliable than the individual models that make up it. One of the big advantages of bagging is its concurrency, since different ensemble models are trained independently of each other. The effectiveness of the proposed approach is shown by the example of predicting the technical state of an object by eight parameters of its functioning. To assess the effectiveness of the application of ensemble machine learning methods for predicting the technical state of an object, the quality criteria of binary classification are used: accuracy, completeness, and F-measure. It is shown that the use of ensemble machine learning methods can improve the accuracy of predicting the state of a technical object by 4% –9% in comparison with basic machine learning methods. This approach can be used by specialists to predict the technical condition of objects in many technical applications, in particular, in aviation.

Handwriting digit classification using PCA-transformed image features

Machine learning model is capable of achieving an incredible great accuracy by training a large amount of data with many dimensions of features in it. To accomplish such a goal, we are currently in desperate need of powerful computers with CPUs (Computation Processing Unit) and GPUs (Graphic Processing Unit) that are competent of the workload of the training process of those machine learning models. Due to the problem that there are multiple disparate computationally intensive tasks that are lengthy period of time during training phase of machine learning, devices like smartphone and personal computer cannot achieve great efficiency during the training phase of machine learning. Instead of enhancing computational performance of devices and improve the machine learning algorithm, another more feasible way to bypass such problem is to apply dimensionality reduction methods on raw data in hope of reducing the number of features that are passing into the machine learning model during the training phase. By doing so, we can shrink the originally large number of features to several dimensions that is accomplishable for our device to compute and meanwhile, sufficient enough to have a model with acceptable accuracy trained. The purpose of this study to explore and investigate the potential benefits and disadvantages of dimensionality reduction on multiple distinct machine learning algorithm by comparing the performances of each machine learning algorithms when passing the features of raw datasets in and when passing the dataset with the features that survived after dimensionality reduction. The performance will be measured by completing a typical machine learning task: handwriting digit classification. Among all the dimensionalities methods that have been discovered, a traditional and conventional dimensionality reduction method, PCA (Principal Components Analysis), is selected. Such method has been proven successful and efficient in generating the set of data with lower dimension features and apply on an image classification task using several supervised machine learning methods, including KNN (k-nearest neighbors), SVM (Support vector machine) and CNN (Convolutional Neural Network).