Jurnal Matematika, Statistika dan Komputasi
Not a member yet
562 research outputs found
Sort by
Performance Evaluation of Classification Methods on Big Data: Decision Trees, Naive Bayes, K-Nearest Neighbors, and Support Vector Machines
Get PDFPerformance evaluation of classification methods on big data is becoming increasingly important in addressing the challenges of data analysis at scale. This study aims to conduct a comparative evaluation of the classification method, namely Decision Trees (DT), Naive Bayes (NB), k-Nearest Neighbors (KNN), and Support Vector Machines (SVM), in analysis on big data evaluated from data simulation and application of real data available in the Rstudio package, namely ISLR. The simulation data used consisted of 2 types of datasets generated based on predictor variables that were normally distributed with different averages and variants and response variables generated in classes adjusted to the characteristics of predictor variables with different proportions. Real data are taken from two types of numeric variables and predictor variables available in the package. The number of sample sizes to be evaluated in each method is n = 500, n = 1000 and n = 5000. In real data, sample division is done randomly to maintain data representativeness. At the evaluation stage, the performance of the method is measured using accuracy metrics. The results of the evaluation of the simulation of Dataset 1 show that the methods that have an influence on the quality of the classification produced if applied to Big Data are the DT and KNN methods. However, in Dataset 2 there is a change in the results of the DT method, because of the influence on the number of classes and the proportion of class distribution in the data. The results obtained from data simulation, proven by applying to real data by showing that similar methods provide a quality influence if applied to Big Data, while the NB and SVM methods do not show a consistent influence when applied to Big Data. The results of observations in this study show that the DT and KNN methods have several advantages that make them suitable for application to Big Data
Implementasi Jump Diffusion Untuk Memprediksi Harga Saham Serta Analisis Risiko Menggunakan Value At Risk Dan Expected Shortfall (Studi Kasus: PT. Indofood Sukses Makmur Tbk)
Get PDFStock prices often fluctuate; therefore, a model is needed to predict the stock price. One of the models that can be used to predict stock prices when experiencing a jump is Jump Diffusion. In addition to predicting, investment is inseparable from the risks that may be borne, so it is also necessary to measure risk. This study aims to implement the Jump Diffusion Model in predicting the stock price of PT Indofood Sukses Makmur Tbk and conduct a risk analysis of the prediction results using Value at Risk (VaR) and Expected Shortfall (ES). In this study, a model was obtained that was used to predict the share price of PT Indofood Sukses Makmur Tbk with a Mean Absolute Percentage Error (MAPE) value of 6.41%. This shows that the accuracy of the stock price prediction results is included in the very good category. In addition, the VaR value of the prediction results with a confidence level of 90%, 95%, and 99% is 0.0292, 0.0372, and 0.0523, and the ES value is 0.0402, 0.0474, and 0.0613.Harga saham sering mengalami fluktuasi oleh karena itu diperlukan suatu model untuk memprediksi harga saham tersebut. Salah satu model yang dapat digunakan untuk memprediksi harga saham saat mengalami lompatan adalah Jump Diffusion. Selain memprediksi, investasi tidak terlepas dari adanya risiko yang mungkin ditanggung sehingga perlu juga untuk dilakukan pengukuran risiko. Penelitian ini bertujuan untuk mengimplementasikan Model Jump Diffusion dalam memprediksi harga saham PT. Indofood Sukses Makmur Tbk, serta melakukan analisis risiko dari hasil prediksi tersebut menggunakan Value at Risk (VaR) dan Expected Shortfall (ES). Pada penelitian ini diperoleh suatu model yang digunakan untuk memprediksi harga saham PT. Indofood Sukses Makmur Tbk dengan nilai Mean Absolute Percentage Error (MAPE) sebesar 6,41%. Hal ini menunjukkan bahwa tingkat keakuratan hasil prediksi harga saham termasuk ke dalam kategori sangat baik. Selain itu, diperoleh Nilai VaR hasil prediksi dengan tingkat kepercayaan 90%, 95%, dan 99% berturut-turut adalah 0,0292; 0,0372; 0,0523 dan nilai ES sebesar 0,0402; 0,0474; 0,0613
Model Machine Learning Stacking untuk Prediksi Pembatalan Pemesanan Hotel
Get PDFThe hotel prepares rooms and resources according to the room booking. Advance booking from customers is a relationship between customers and hotels that ensures price stability for customers to enjoy services. Cancellation of hotel bookings and inability to satisfy potential customers is a widespread and alarming problem that can increase hotel operating costs and affect customer satisfaction. Given that the impact on the hospitality industry can be very bad, predicting hotel cancellations can be a solution to help build an appropriate operational strategy. Method used in this research is stacking machine learning model. Stacking consists of two levels, where in this study level 0 (base learner) uses the Naive Bayes, Logistic Regression, and Gradient Boosting Machine algorithms while at level 1 (meta learner) uses the Random Forest algorithm. Accuracy value of the stacking model classification and the gradient boosting machine has the highest accuracy value of 0.87. Sensitivity value of the stacking model is 0.86 and is the highest sensitivity value which means that the stacking model classification is very precise in predicting consumers in canceling hotel reservations. Specificity value of the gradient boosting machine is 0.88 and is the highest specificity value, which means that the gradient boosting machine classification is very precise in predicting consumers who do not cancel hotel reservations. Naive bayes and logistic regression classifications have accuracy, sensitivity, specificity, precision values that are not high.
Hotel mempersiapkan kamar dan sumber daya sesuai dengan pemesanan kamar. Pemesanan di awal dari pelanggan merupakan hubungan antara pelanggan dengan hotel yang memastikan kestabilan harga bagi pelanggan untuk menikmati layanan. Pembatalan pemesanan hotel dan ketidakmampuan untuk memuaskan calon konsumen merupakan masalah yang meluas dan mengkhawatirkan yang dapat meningkatkan biaya operasional hotel dan mempengaruhi kepuasan pelanggan. Mengingat hal itu dampaknya terhadap industri perhotelan bisa sangat buruk, maka dengan memprediksi pembatalan hotel dapat menjadi solusi untuk membantu membangun strategi operasional yang sesuai. Metode yang digunakan pada penelitian ini adalah stacking machine learning model. Stacking terdiri dari dua level, dimana pada penelitian ini level 0 (base learner) menggunakan algoritma Naive Bayes, Logistic Regression, dan Gradient Boosting Machine sedangkan pada level 1 (meta learner) menggunakan algoritma Random Forest. nilai akurasi klasifikasi stacking model dan gradient boosting machine memiliki nilai akurasi tertinggi sebesar 0.87. Nilai sensitivitas stacking model sebesar 0.86 dan merupakan nilai sensitivitas tertinggi yang berarti klasifikasi stacking model sangat tepat memprediksi konsumen dalam pembatalan pemesanan hotel. Nilai Spesifisitas gradient boosting machine sebesar 0.88 dan merupakan nilai spesifisitas tertinggi yang berarti klasifikasi gradient boosting machine sangat tepat memprediksi konsumen yang tidak melakukan pembatalan pemesanan hotel. Klasifikasi naive bayes dan logistic regression memiliki nilai akurasi, sensitivitas, spesifisitas, presisi yang tidak tinggi.
Kata kunci: Stacking Model, Base Learner, Meta Learne
Optimasi Portofolio Saham Indeks Bisnis 27 Menggunakan Model Black Litterman Disertai Perhitungan Value At Risk: Bahasa Indonesia
No full textAn investment can provide a profit with a certain level of risk for an investor both now and in the future. This indicates that investments are important in both financial and asset management. Finance investments can be made on several stocks or portfolios. To profit from an investment, you need a tool to optimize profit and risk, which is a portfolio. This research uses the Black Litterman Model in portfolio optimization along with Value at Risk (VaR) calculations to determine the risk of each stock. The data used is close price data on the Business Index 27 for the period January-December 2023. Next, selected 9 shares in the formation of an optimal portfolio namely, AKRA, AMRT, ASII, BBCA, BBNI, BBRI, INKP, KLBF and TLKM. Based on the calculations, the rate of profit achieved on the portfolio is 5.48% with a risk of 0.40%. Then use the Historical Method and the Monte Carlo Simulation Method to calculate the VaR using nine optimal stocks, with a 95% confidence rate. In the Monte Carlo simulation, 300 repetitions of VaR calculations are performed. Different results on both methods are due to different approaches to risk calculationInvestasi dapat memberikan keuntungan disertai tingkat risiko tertentu bagi seorang investor baik saat ini maupun di masa depan. Hal ini menandakan bahwa investasi merupakan hal penting dalam pengelolaan keuangan maupun aset. Investasi dalam keuangan dapat dilakukan pada beberapa saham atau portofolio. Untuk memperoleh keuntungan pada investasi diperlukan suatu alat untuk mengoptimalkan keuntungan serta risiko, yaitu portofolio. Penelitian ini menggunakan Model Black Litterman dalam mengoptimalkan portofolio disertai perhitungan Value at Risk (VaR) untuk menentukan risiko masing-masing saham. Data yang digunakan merupakan data close price pada Indeks Bisnis 27 periode Januari-Desember 2023. Selanjutnya, dipilih 9 saham dalam pembentukan portofolio optimal yaitu, AKRA, AMRT, ASII, BBCA, BBNI, BBRI, INKP, KLBF dan TLKM. Berdasarkan hasil perhitungan, tingkat keuntungan yang didapatkan pada portofolio tersebut yaitu sebesar 5.48% dengan risiko sebesar 0.40%. Selanjutnya digunakan Metode Historis dan Metode Simulasi Monte Carlo untuk menghitung VaR menggunakan 9 saham optimal, dengan tingkat kepercayaan 95%. Dalam Simulasi Monte Carlo, dilakukan 300 kali pengulangan perhitungan VaR. Hasil yang berbeda pada kedua metode tersebut disebabkan oleh pendekatan yang berbeda dalam menghitung risiko
The Comparison of Inverse Gaussian and Gamma Regression: Application on Stunting Data in Jepara
Get PDFMany research data have distributions other than the normal distribution, called exponential family distributions. The exponential family of distributions includes the inverse Gaussian and Gamma distributions. There are parallels between these two distributions in terms of the kind of random variable and how well they work. Finding the optimal model using inverse Gaussian and Gamma regression on stunting data in Jepara is the goal of this study. Maximum Likelihood Estimation is used for parameter estimation, Maximum Likelihood Ratio Test is used for simultaneous parameter testing, and Wald testing is used for partial parameter testing. For this case, the best model is inverse Gaussian regression. Exclusive breastfeeding, low birth weight babies, clean drinking water facilities, and the number of Integrated Service Post (Posyandu) influence the percentage of stunting in Jepara.
Perbandingan Metode Klasifikasi dalam Memprediksi Penjualan Produk Ban Terlaris
No full textData mining is a term to describe the process of moving through large databases in search of certain previously unknown patterns. In finding certain patterns, you need a supporting technique, called machine learning. Machine learning involves learning hidden patterns in data and further using patterns to classify or predict an event related to a problem. One of the problems can be solved with machine learning such as predicting the sales rate of tire products. This can help companies predict tire products that are selling well in the market. In producing an accurate prediction model, it will be compared with decision tree classification methods of CART, CART + Discrete Adaboost, and Naive Bayes applied to tire sales data by PT. Mitra Mekar Mandiri. The results of the study based on successive model performance evaluations are model Naive Bayes < model CART < model CART+Discrete Adaboost. The Discrete Adaboost model with a data proportion of 90:10 is the best model for predicting tire sales. The accuracy, sensitivity and specificity values for the model were 79.17%; 89.47%; and 68.84%. The AUC value is 0.8 which indicates the model is goodData mining merupakan istilah untuk mendeskripsikan proses perpindahan melalui database besar untuk mencari pola tertentu yang sebelumnya tidak diketahui. Dalam menemukan pola tertentu maka perlu teknik yang mendukung yang disebut machine learning. Machine learning melibatkan pembelajaran pola tersembunyi dalam data dan selanjutnya menggunakan pola untuk mengklasifikasikan atau memprediksi suatu peristiwa yang terkait dengan masalah. Penelitian bertujuan untuk membandingkan metode klasifikasi decision tree tipe CART, discrete adaboost, dan naive bayes yang diaplikasikan pada data penjualan ban oleh PT. Mitra Mekar Mandiri. Hasil penelitian berdasarkan evaluasi kinerja model berturut-turut adalah model discrete adaboost < model naive bayes < model CART. Model CART dengan proporsi data 80:20 adalah model terbaik dalam memprediksi penjualan ban terlaris. Nilai akurasi, sensitifitas dan spesifisitas untuk model tersebut masing-masing 77,5%; 73,5%; dan 81,3%. Nilai AUC diperoleh 0,774 yang mengindikasikan model cukup bai
Estimasi Conditional Value at Risk Pada Perusahaan Sektor Consumer Non-Cyclicals Menggunakan Pendekatan Extreme Value Theory
Get PDFConditional Value at Risk (CVaR) is an estimate of the risk of loss that exceeds the Value at Risk (VaR) level. VaR is one of the most commonly used stock risk measurement methods to assess the risk of large investments. Extreme Value Theory (EVT) is a method used to analyze data that contains extreme values. The goal of EVT is to estimate the probability of an extreme event occurring by examining the tails of a distribution based on observed extreme values. There are two general distributions used in EVT, namely Generalized Extreme Value (GEV) and Generalized Pareto Distribution (GPD). This research aims to determine the estimated level of loss that investors may experience when investing in PT Hanjaya Mandala Sampoerna Tbk (HMSP) and PT Japfa Comfeed Indonesia Tbk (JPFA). The L-Moment method is applied to estimate the parameters in this distribution so that an explicit parameter form is obtained. Based on CVaR analysis using the Block Maxima (BM) approach, investors in HMSP and JPFA are estimated to experience losses of 20.0752% and 29.6537% respectively. Using the Peaks Over Threshold (POT) approach, the estimated losses are 0.966% and 1.548% for HMSP and JPFA, respectively. Based on CVaR calculations using both approaches, the POT approach with GPD provides a more accurate and reliable investment risk estimate than the BM approach with GEV distributionConditional Value at Risk (CVaR) merupakan perkiraan risiko kerugian yang melebihi tingkat Value at Risk (VaR). VaR adalah salah satu metode pengukuran risiko saham yang paling umum digunakan untuk menilai risiko investasi besar. Extreme Value Theory (EVT) adalah metode yang digunakan untuk menganalisis data yang mengandung nilai ekstrem. Tujuan EVT adalah memperkirakan kemungkinan terjadinya peristiwa ekstrem dengan memeriksa ujung distribusi berdasarkan nilai ekstrem yang diamati. Terdapat dua distribusi umum yang digunakan dalam EVT yaitu Generalized Extreme Value (GEV) dan Generalized Pareto Distribution (GPD). Penelitian ini bertujuan untuk mengetahui perkiraan tingkat kerugian yang mungkin dialami investor ketika berinvestasi pada PT Hanjaya Mandala Sampoerna Tbk (HMSP) dan PT Japfa Comfeed Indonesia Tbk (JPFA). Metode L-Moment diterapkan untuk mengestimasi parameter pada distribusi tersebut sehingga diperoleh bentuk parameter yang eksplisit. Berdasarkan analisis CVaR dengan pendekatan Block Maxima (BM), investor pada HMSP dan JPFA diperkirakan mengalami kerugian masing-masing sebesar 20,0752% dan 29,6537%. Dengan menggunakan pendekatan Peaks Over Threshold (POT), estimasi kerugian masing-masing sebesar 0,966% dan 1,548% untuk HMSP dan JPFA. Berdasarkan perhitungan CVaR menggunakan kedua pendekatan tersebut, pendekatan POT dengan GPD memberikan perkiraan risiko investasi yang lebih akurat dan andal dibandingkan pendekatan BM dengan distribusi GEV
Agglomerative Nesting Cluster Analyst in Mapping District/City Health Facilities in West Java Province
Get PDFThe use of Hierarchical Clustering is used to group districts or cities in West Java according to the number of health facilities, distance to health facilities and population density using Agglomerative Nesting (AGNES). Clustering in this study utilizes complete linkage clustering. The elbow method produces two optimal clusters which are then validated with the sillhoute coefficient and Calinski-Harabasz. In this study, there are 27 variables in the form of health facilities spread across 27 regencies/cities in West Java in 2021. The results of the cluster analysis formed in this study are 18 districts/cities in cluster one and 9 districts/cities in cluster twoPemanfaatan Hierarchical Clustering digunakan untuk mengelompokkan kabupaten atau kota di Jawa Barat menurut jumlah sarana Kesehatan, jarak menuju sarana Kesehatan dan kepadatan penduduk menggunakan Agglomerative Nesting (AGNES). Clustering pada penelitian ini memanfaatkan clustering complete linkage. Metode elbow menghasilkan dua cluster optimal yang kemudian divalidasi dengan sillhoute coefficient dan Calinski-Harabasz. Pada penelitian ini terdapat 27 peubah berupa sarana kesehatan yang tersebar pada 27 kabupaten/kota di Jawa Barat tahun 2021. Hasil analisis cluster yang terbentuk dalam penelitian ini terdapat 18 kabupaten/kota pada cluster satu dan 9 kabupaten/kota pada cluster dua
Spatial Weighting Selection in GSTAR and S-GSTAR Models for Temperature Prediction
Get PDFRecent research in time series analysis indicates that events at a particular location are not only influenced by events at previous times but also by proximity between locations. Events influenced by both space and time can be modeled using a space-time model. GSTAR model is one such space-time model. In its development, time series data exhibiting seasonal patterns are modeled using Seasonal GSTAR (S-GSTAR). The GSTAR and S-GSTAR models are used to model temperature in the Banjar, Cilacap, and Sleman Districts. The purpose of employing both methods is to compare the best model for modeling temperature at these three locations. Spatial weights used include inverse distance weighting using the Euclidean distance formula, uniform weighting, and cross-correlation normalization weighting. Ordinary Least Squares (OLS) is the estimation method used in this study. The best model obtained is S-GSTAR with inverse distance weighting, as this model has the smallest RMSE value.Penelitian terkini dalam deret waktu menunjukkan bahwa kejadian pada suatu lokasi bukan hanya dipengaruhi kejadian pada waktu sebelumnya, tetapi juga kedekatan antarlokasi. Kejadian yang dipengaruhi oleh ruang dan waktu dapat dimodelkan dengan model ruang waktu. Contoh dari model ruang waktu adalah model GSTAR. Pada perkembangannya, untuk data ruang waktu yang memiliki pola musiman dapat dimodelkan dengan Seasonal GSTAR (S-GSTAR). Model GSTAR dan S-GSTAR digunakan untuk memodelkan temperatur di Kabupaten Banjar, Cilacap, Sleman. Tujuan penggunaan kedua metode tersebut adalah untuk membandingkan model yang terbaik dalam memodelkan temperature pada ketiga lokasi tersebut. Bobot spasial yang digunakan yaitu bobot invers jarak dengan rumus euclidian distance, bobot seragam, dan bobot normalisasi korelasi silang. OLS (Ordinary Least Square) merupakan metode estimasi yang digunakan pada penelitian ini. Model terbaik yaitu model S-GSTAR dengan bobot spasial inversi jarak, hal ini dikarenakan model tersebut nilai RMSE terkecil.
 
Penentuan Harga Opsi Saham dengan Menggunakan Binomial Trees dengan Penyertaan Implied Volatility
Get PDFThe Black-Scholes model provides an analytical solution in option pricing and has been widely used in finance. This model assumes constant volatility. Pricing option incorporating implied volatility is conducted using implied binomial tree. This study aims to simulate the prices of put options and call options using implied binomial trees, binomial trees and the Black-Scholes model and determine the factors that influence option prices. The simulation was conducted using Matlab. The option price resulted from implied binomial tree and binomial tree are compared with the option prices of the Black-Scholes model to determine the difference of option prices with constant volatility and option prices incorporating implied volatility. The implied binomial tree method provides better option prices than the binomial tree based on small relative error value to the Black-Scholes model. This is caused by the transition probability value of stock price movements in the implied binomial tree at each point is different, whereas in the binomial tree the value of transition probability is same. Furthermore, increasing the time step causes the option prices obtained from the implied binomial tree converge to the Black-Scholes. It is concluded that these three methods can be used in option pricing. Factors that influence the option price are stock price, strike price, interest rate and maturity date, are also obtainedModel Black-Scholes memberikan solusi analitik pada penentuan harga opsi dan telah digunakan secara luas dalam dunia keuangan. Namun, asumsi volatilitas konstan pada model Black-Scholes kurang merepresentasikan kondisi riil. Model Black-Scholes tidak dapat lagi menjelaskan implied volatility di pasar opsi. Pada penelitian ini harga opsi dimodelkan dengan implied binomial tree, yaitu metode binomial tree yang unsur volatilitasnya menggunakan efek implied volatility yang lebih konsisten terhadap kondisi riil. Selanjutnya dilakukan simulasi harga opsi jual dan opsi beli menggunakan metode binomial tree standar dan implied binomial tree. Hasil simulasi menunjukkan harga opsi konsisten dengan harga opsi model Black-Scholes. Berdasarkan simulasi, diperolah pula bahwa metode implied binomial tree memberikan harga opsi yang lebih baik daripada binomial tree standar berdasarkan nilai eror yang dihasilkan. Lebih lanjut, peningkatan langkah waktu mengakibatkan harga opsi yang diperoleh dari implied binomial tree konvergen ke harga opsi dari model Black-Scholes. Selain itu, diperoleh faktor-faktor yang berpengaruh terhadap harga opsi yaitu harga saham, harga strike, suku bunga, dan waktu jatuh tempo.