scholarly journals Features of building a predictive neuro-fuzzy network

2020 ◽  
pp. 13-17
Author(s):  
А.И. Епихин ◽  
Е.В. Хекерт ◽  
А.Б. Каракаев ◽  
М.А. Модина
Keyword(s):  
Mean Squared Error ◽  
Least Squares Method ◽  
Finite Impulse Response ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Piston Engine ◽  
Absolute Percentage Error ◽  
Squared Error ◽  
Neuro Fuzzy ◽  
Fuzzy Network

В статье рассматриваются особенности построения прогностической нейро-фаззи сети. В процессе исследования представлена структура адаптивного нейро-фаззи-предиктора и многомерного нейро-фаззи-нейрона. Рассмотрен принцип обработки информации, поступающей в режиме реального времени, о работе поршневого двигателя СЭУ с использованием TSK-системы нулевого порядка с применением быстродействующих оптимизационных процедур второго порядка типа рекуррентного метода наименьших квадратов для настройки синаптических весов. Определена архитектура искусственной нейро-фаззи сети для прогноза ресурсной прочности поршневого двигателя СЭУ марки RND 105, состоящая из пяти последовательно соединенных слоев. Представлена структура динамических нейронов-фильтров с конечной импульсной характеристикой. Рассмотрена процедура обучения нейросети. При проведения численного эксперимента использовались следующие критерии оценки: MSE (mean squared error, среднеквадратичная погрешность); SMAPE (Symmetric mean absolute percentage error, симметрично абсолютная процентная погрешность) - характеризует погрешность прогноза в процентах. Экспериментальный анализ разработанной сети проводился на примере прогнозирования ресурсной прочности восьмицилиндрового двухтактного судового дизеля марки RND 105. The article discusses the features of building a predictive neuro-fuzzy network. During the research, the structure of an adaptive neuro-fuzzy predictor and a multidimensional neuro-fuzzy neuron is presented. The principle of processing information received in real time about the operation of a piston engine of a SEP using a TSK-system of zero order with the use of high-speed optimization procedures of the second order such as the recurrent least squares method for adjusting synaptic weights is considered. The architecture of an artificial neuro-fuzzy network for predicting the resource strength of a piston engine SEU brand RND 105, consisting of five layers connected in series, has been determined. The structure of dynamic filter neurons with finite impulse response is presented. The procedure for training a neural network is considered. During the numerical experiment, the following evaluation criteria were used: MSE (mean squared error); SMAPE (Symmetric mean absolute percentage error) - characterizes the forecast error in percentage. An experimental analysis of the developed network was carried out on the example of predicting the resource strength of an eight-cylinder two-stroke marine diesel engine of the RND 105 brand.

scholarly journals Peramalan Jumlah Populasi Sapi Potong di Kalimantan Selatan Menggunakan Metode Moving Average, Exponential Smoothing dan Trend Analysis

2019 ◽  
Vol 6 (1) ◽  
pp. 41
Author(s):  
Jaka Darma Jaya
Keyword(s):  
Trend Analysis ◽  
Mean Squared Error ◽  
Moving Average ◽  
Exponential Smoothing ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Absolute Percentage Error ◽  
Polynomial Trend ◽  
Squared Error

Perkembangan produksi daging sapi di Indonesia selama 30 tahun terakhir secara umum cenderung meningkat. Kebutuhan daging sapi di Indonesia masih belum bisa dicukupi oleh supply domestik, sehingga diperlukan impor daging sapi dari luar negeri.  Diperlukan kajian tentang proyeksi ketersediaan populasi sapi potong di masa mendatang agar diambil kebijakan yang tepat dalam menjaga stabilitas dan keterpenuhan supply daging nasional.  Penelitian ini bertujuan untuk melakukan peramalan jumlah populasi sapi potong menggunakan 3 (tiga) metode peramalan yaitu metode moving average, exponential smoothing dan trend analysis.  Hasil peramalan ini selanjutnya diukur akurasinya menggunakan MAD (Mean Absolud Deviation), MSE (Mean Squared Error) dan MAPE (Mean Absolute Percentage Error).  Proyeksi populasi sapi potong pada tahun 2019 (periode berikutnya) menggunakan 3 metode peramalan adalah: 195.100 (moving average); 218.225 (exponential smooting) dan 262.899 (trend analysis). Pengukuran akurasi menggunakan MAD, MSE dan MAPE menunjukkan bahwa metode peramalan jumlah populasi sapi potong yang paling akurat adalah peramalan menggunakan metode polynomial trend analysis (MAD 14.716,12;  MSE 327.282.084,17; dan MAPE 0,09) karena memiliki tingkat kesalahan yang lebih kecil dibandingkan hasil peramalan menggunakan metode moving average dan exponential smoothing.

scholarly journals PERBANDINGAN METODE REGRESI LINEAR DAN NEURAL NETWORK BACKPROPAGATION DALAM PREDIKSI NILAI UJIAN NASIONAL SISWA SMP MENGGUNAKAN SOFTWARE R

JOUTICA ◽  
2020 ◽  
Vol 5 (1) ◽  
pp. 331
Author(s):  
Masruroh Masruroh
Keyword(s):  
Neural Network ◽  
Mean Squared Error ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Output Data ◽  
Root Mean Squared Error ◽  
Absolute Percentage Error ◽  
Squared Error

Metode regresi linear dan neural network backpropagation merupakan metode yang kerap digunakan dalam model prediksi. Penelitian ini bertujuan untuk membandingkan akurasi metode regresi linear dan backpropagation dalam prediksi nilai Ujian Nasional siswa SMP. Data yang digunakan berupa data nilai ujian akhir semester dan ujian sekolah sebagai input dan nilai ujian nasional sebagai output. Data didapatkan dari SMPN 1 dan SMPN 2 Lamongan.. Jumlah dataset sebanyak 701 dibagi menjadi 75% data training dan 25% data testing. Simulasi prediksi dilakukan menggunakan software R. Parameter akurasi yang digunakan adalah Root Mean Squared Error (RMSE) dan Mean Absolute Percentage Error (MAPE). Hasil penelitian menunjukkan model prediksi menggunakan metode regresi linear menghasilkan RMSE sebesar 9,04 dan MAPE sebesar 3,94%, sedangkan model prediksi menggunakan backpropagation menghasilkan RMSE sebesar 7,28 dan MAPE sebesar 0,55%. Dengan demikian dalam penelitian ini metode neural network backpropagation memiliki akurasi yang lebih baik dalam prediksi nilai Ujian Nasional siswa SMP.

scholarly journals Rainfall Analysis and Forecasting Using Deep Learning Technique

2021 ◽  
Vol 2 (2) ◽  
pp. 1-11
Author(s):  
Pragati Kanchan ◽  
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Mean Squared Error ◽  
Predictive Accuracy ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Root Mean Squared Error ◽  
Rainfall Prediction ◽  
Absolute Percentage Error ◽  
Squared Error

Rainfall forecasting is very challenging due to its uncertain nature and dynamic climate change. It's always been a challenging task for meteorologists. In various papers for rainfall prediction, different Data Mining and Machine Learning (ML) techniques have been used. These techniques show better predictive accuracy. A deep learning approach has been used in this study to analyze the rainfall data of the Karnataka Subdivision. Three deep learning methods have been used for prediction such as Artificial Neural Network (ANN) - Feed Forward Neural Network, Simple Recurrent Neural Network (RNN), and the Long Short-Term Memory (LSTM) optimized RNN Technique. In this paper, a comparative study of these three techniques for monthly rainfall prediction has been given and the prediction performance of these three techniques has been evaluated using the Mean Absolute Percentage Error (MAPE%) and a Root Mean Squared Error (RMSE%). The results show that the LSTM Model shows better performance as compared to ANN and RNN for Prediction. The LSTM model shows better performance with mini-mum Mean Absolute Percentage Error (MAPE%) and Root Mean Squared Error (RMSE%).

scholarly journals Analisis Error Terhadap Peramalan Data Penjualan

Techno Com ◽  
2021 ◽  
Vol 20 (1) ◽  
pp. 1-9
Author(s):  
Alyauma Hajjah ◽  
Yulvia Nora Marlim
Keyword(s):  
Mean Squared Error ◽  
Moving Average ◽  
Exponential Smoothing ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Absolute Percentage Error ◽  
Squared Error

Tujuan penelitian ini membahas tentang peramalan permintaan lampu LED bermerk Sanyo. Penelitian ini menggunakan metode Moving Average dan Exponential Smoothing. Pada Metode Moving Average digunakan untuk peramalan periode 3 bulan dan 5 bulan, sedangkan metode Exponential Smoothing menggunakan parameter  = 0,1; 0,5; 0,7 dan 0,9. Dari hasil peramalan setiap metode dibandingkan nilai dari error, adapun nilai error yang dibahas adalah Mean Absolut Deviation (MAD), Mean Squared Error (MSE), dan Mean Absolute Percentage Error (MAPE). Berdasarkan hasil penelitian ini metode yang memiliki tingkat kesalahan terkecil adalah metode Exponential Smoothing dengan parameter , yang memiliki nilai MAD= 1.214,54; MSE = 2.758.993 dan MAPE = 9,17%. Dapat disimpulkan bahwa metode yang paling optimal digunakan untuk meramalkan jumlah permintaan lampu pada bulan berikutnya adalah metode Exponential Smoothing dengan parameter . Hasil peramalan permintaan lampu pada bulan Januari 2020 dengan menggunakan metode Exponential Smoothing dengan  adalah 15.800,88. Sehingga untuk peramalan permintaan untuk bulan berikutnya peneliti merekomendasikan menggunakan metode Exponential Smoothing dengan perameter .

scholarly journals Forecast Error Calculation with Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE)

2020 ◽  
Vol 1 (2) ◽  
Author(s):  
Ansari Saleh Ahmar
Keyword(s):  
Decision Making ◽  
Mean Squared Error ◽  
Forecast Error ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Forecasting Accuracy ◽  
Error Calculation ◽  
Absolute Percentage Error ◽  
Squared Error ◽  
High Level

Calculation errors in forecasting a data are very important from a forecasting process. The high level of forecasting accuracy will affect the level of confidence in forecasting decision making.

scholarly journals Aplikasi Triple Exponential Smoothing Untuk Forecasting Jumlah Penduduk Miskin

2020 ◽  
Vol 3 (2) ◽  
pp. 76-82
Author(s):  
Padrul Jana
Keyword(s):  
Poverty Reduction ◽  
Mean Squared Error ◽  
Exponential Smoothing ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Poor People ◽  
Mean Absolute Deviation ◽  
Absolute Deviation ◽  
Absolute Percentage Error ◽  
The Government

This study aims to predict the number of poor in Indonesia for the next few years using a triple exponential smoothing method.The purpose of this research is the result of the forecast number of poor people in Indonesia accurate forecast results are used as an alternative data the government for consideration of government to determine the direction of national poverty reduction policies. This research includes the study of literature research, by applying the theory of forecasting to generate predictions of poor people for coming year. Furthermore, analyzing the mistakes of the methods used in terms of the count: Mean Absolute Deviation (MAD), Mean Square Error (MSE), Mean absolute percentage error (MAPE) and Mean Percentage Error (MPE). The function of this error analysis is to measure the accuracy of forecasting results that have been conducted.These results indicate that the number of poor people in 2017 amounted to 24,741,871 inhabitants, in 2018 amounted to 24,702,928 inhabitants, in 2019 amounted to 24,638,022 inhabitants and in 2020 amounted to 24,547,155 people. The forecasting results show an average reduction in the number of poor people in Indonesia last five years (2016-2020 years) ranges from 0.16 million. Analysis forecasting model obtained an mean absolute deviation (MAD) obtained by 0.246047. Mean squared error (MSE) of forecasting results with the original data by 1.693277. Mean absolute percentage error (MAPE) of 3.040307% and the final Mean percentage error (MPE) of 0.888134%.Kata Kunci: Forecasting, Triple Exponential Smoothing

scholarly journals ARDL METHOD: Forecasting Data Curah Hujan Harian NTB

Jurnal Varian ◽  
2020 ◽  
Vol 3 (2) ◽  
pp. 73-82
Author(s):  
Ulul Azmi ◽  
Zilullah Nazir Hadi ◽  
Siti Soraya
Keyword(s):  
Mean Squared Error ◽  
Percentage Error ◽  
Mean Absolute Deviation ◽  
Absolute Deviation ◽  
Root Mean Squared Error ◽  
Absolute Percentage Error ◽  
Distributed Lag ◽  
Autoregressive Distributed Lag ◽  
Squared Error ◽  
Parameter Error

Penelitian ini berisi tentang prediksi atau forecasting data iklim di Nusa Tenggara Barat (NTB) tahun 2011, yakni jumlah hari terjadinya hujan dengan menggunakan metode Autoregressive Distributed Lag (ARDL). Data yang digunakan yaitu data iklim di Nusa Tenggara Barat (NTB) dari tahun 2006 -2010, dengan menggunakan beberapa parameter error seperti Mean Absolute Deviation (MAD), Mean Squared Error (MSE), Root Mean Squared Error (RMSE) dan Mean Absolute Percentage Error (MAPE). Berdasarkan hasil simulasi data iklim di Nusa Tenggara Barat (NTB) tersebut, diperoleh prediksi jumlah hari terjadinya curah hujan pada tahun 2011 sebesar 226 hari dengan nilai MAD 20,8069, MSE 3,5569, RMSE 1,88597, dan MAPE 11,9297 . Dan prediksi jumlah hari terjadinya hujan pada tahun 2011 sebanyak 225,928 hari atau jika di bulatkan menjadi 226 hari dengan nilai parameter error MAD sebesar 20,8069, sehingga dapat disimpulkan pada tahun 2011 terjadi peningkatan jumlah hari terjadinya hujan di Nusa Tenggara Barat (NTB).

scholarly journals Previsão dos modelos univariados e rede neural da demanda turística do estado de Mato Grosso do Sul

2020 ◽  
Vol 19 (3) ◽  
Author(s):  
Bruno Matos Porto ◽  
Daniela Althoff Philippi ◽  
Vanessa Aline Wagner Leite
Keyword(s):  
Mean Squared Error ◽  
Percentage Error ◽  
Mean Absolute Deviation ◽  
Absolute Deviation ◽  
Mato Grosso ◽  
Absolute Percentage Error ◽  
Squared Error ◽  
Mato Grosso Do Sul

O objetivo deste artigo foi gerar previsões de curto, médio e longo prazos e comparar a precisão dos modelos em cada horizonte de previsão. Para atender o objetivo foram aplicados os modelos univariados e rede neural (NNAR) nos dados da demanda turística do estado de Mato Grosso do Sul (MS). A amostra foi coletada na ferramenta base de dados extrator do Instituto Brasileiro de Turismo (Embratur) referente as chegadas turísticas por todas as vias registradas no MS entre janeiro de 2007 a dezembro de 2017. As previsões dos modelos de previsão ARIMA, Holt-Winters (HW) versões aditiva e multiplicativa e NNAR foram projetadas, por meio da linguagem de programação R, com uso do software R Studio. O procedimento empírico de execução dos scripts de todos os modelos foi disponibilizado. As predições fora da amostra da procura do turismo abrangeram o intervalo de janeiro até dezembro de 2018, sendo então comparadas aos dados reais do mesmo período. As previsões dos modelos foram comparadas no curto, médio e longo prazo mediante os critérios Mea Absolute Percentage Error (MAPE), Mean Absolute Deviation (MAD) e Mean Squared Error (MSD). A rede neural (NNAR) superou os modelos testados em diferentes horizontes de previsão e as medidas de erros mostraram que a NNAR é altamente precisa. Em segundo lugar no ranking de acuracidade destacou-se ARIMA. Os resultados mostraram que as previsões da rede neural auxiliam na tomada de decisão dos planejadores turísticos de MS. Para pesquisas futuras recomenda-se realizar previsões fora da amostra num amplo número de séries temporais.

Rancang Bangun Aplikasi Peramalan Laba dengan Metode Kuadrat Terkecil Berbasis Android

2016 ◽  
Vol 7 (2) ◽  
pp. 125-130
Author(s):  
Audrey Sugiarto ◽  
Seng Hansun
Keyword(s):  
Least Squares ◽  
Least Squares Method ◽  
The State ◽  
Mean Absolute Percentage Error ◽  
Percentage Error ◽  
Absolute Percentage Error ◽  
Android Os ◽  
The Future ◽  
Using Data ◽  
Technology Effects

The advancement of technology effects in increasing competition between companies. Because of that, companies need more than just raw information, but rather some insight that can help companies to make decisions in the future regarding all the possibilities that can happen. The data that can help the company to make decisions is a forecasting earnings because it can help predict the state of the company has right now, and also can help to make a better decision in the future. Therefore, this study discusses about the design and development of forecasting earnings application using Least Squares Method which will create an equation with the formula, y = ax + b. The method will be implemented based on Android OS at PT TRI PANJI GEMILANG using data from January 2005 to December 2013 (108 months) for data forecasting, and the data used to check the error is data from January 2014 to May 2015 (17 months). Forecasting results have a mean absolute percentage error (MAPE) about 8.26%, with an accuracy of forecasting results about 91.74%. Keywords: android, forecasting, least squares method, profits

Sign in / Sign up

Export Citation Format

Share Document