scholarly journals Modelling the Working Week for Multi-Step Forecasting using Gaussian Process Regression

2017 ◽  
Author(s):  
Pasan Karunaratne ◽  
Masud Moshtaghi ◽  
Shanika Karunasekera ◽  
Aaron Harwood ◽  
Trevor Cohn
Keyword(s):  
Time Series ◽  
Gaussian Process ◽  
Real World ◽  
Time Series Data ◽  
Gaussian Process Regression ◽  
Series Data ◽  
Combination Methods ◽  
The University ◽  
The City ◽  
Source Of Information

In time-series forecasting, regression is a popular method, with Gaussian Process Regression widely held to be the state of the art. The versatility of Gaussian Processes has led to them being used in many varied application domains. However, though many real-world applications involve data which follows a working-week structure, where weekends exhibit substantially different behavior to weekdays, methods for explicit modelling of working-week effects in Gaussian Process Regression models have not been proposed. Not explicitly modelling the working week fails to incorporate a significant source of information which can be invaluable in forecasting scenarios. In this work we provide novel kernel-combination methods to explicitly model working-week effects in time-series data for more accurate predictions using Gaussian Process Regression. Further, we demonstrate that prediction accuracy can be improved by constraining the non-convex optimization process of finding optimal hyperparameter values. We validate the effectiveness of our methods by performing multi-step prediction on two real-world publicly available time-series datasets - one relating to electricity Smart Meter data of the University of Melbourne, and the other relating to the counts of pedestrians in the City of Melbourne.

scholarly journals ANALISIS PENGARUH INVESTASI SWASTA , PENGELUARAN PEMERINTAH, DAN PENDUDUK TERHADAP PERTUMBUHAN EKONOMI DI KOTA BITUNG

2019 ◽  
Vol 19 (1) ◽  
Author(s):  
Yandiles Weya ◽  
Vecky A.J. Masinambow ◽  
Rosalina A.M. Koleangan
Keyword(s):  
Economic Growth ◽  
Time Series ◽  
Economic Activity ◽  
Time Series Data ◽  
Private Investment ◽  
Least Square ◽  
Series Data ◽  
Ordinary Least Square ◽  
North Sulawesi ◽  
The City

ANALISIS PENGARUH INVESTASI SWASTA , PENGELUARAN PEMERINTAH, DAN PENDUDUK TERHADAP PERTUMBUHAN EKONOMI DI KOTA BITUNG Yandiles Weya, Vecky A.J. Masinambow, Rosalina A.M. Koleangan. Fakultas Ekonomi dan Bisnis, Magister Ilmu EkonomiUniversitas Sam Ratulangi, Manado ABSTRAKPada suatu periode perekonomian mengalami pertumbuhan negatif berarti kegiatan ekonomi pada periode tersebut mengalami penurunan. Kota Bitung periode tahun 2004-2014 mengalami pertumbuhan ekonomi yang fluktuasi. Adanya fluktuasi ini dapat dipengaruhi oleh investasi swasta, belanja langsung, dan penduduk Pertumbuhan ekonomi merupakan salah satu tolok ukur keberhasilan pembangunan ekonomi di suatu daerah. Pertumbuhan ekonomi mencerminkan kegiatan ekonomi. Pertumbuhan ekonomi dapat bernilai positif dan dapat pula bernilai negatif. Jika pada suatu periode perekonomian mengalami pertumbuhan positif berarti kegiatan ekonomi pada periode tersebut mengalami peningkatan. Sedangkan jikaTahun 2004-2014 yang bersumber dari Badan Pusat Statistik Provinsi Sulut dan Kota Bitung. Metode analisis yang digunakan adalah model ekonometrik regresi berganda double-log (log-log) dengan metode Ordinary Least Square (OLS). Penelitian ini bertujuan untuk mengetahui apakah perkembangan investasi swasta, belanja langsung, dan penduduk berpengaruh terhadap pertumbuhan ekonomi Kota Bitung. Data yang dipakai menggunakan data time series periodeHasil regresi model pertumbuhan ekonomi dengan persamaan regresinya yaitu  LPDRB  =  - 4,445    +  0.036 LINV  +  0.049 LBL  +  2,229 LPOP.  Dari hasil tersebutmenunjukkan perkembangan investasi swasta, belanja langsung dan penduduk berpengaruh positif dan signifikan terhadap pertumbuhan ekonomi Kota Bitung.Kata Kunci :pertumbuhan ekonomi, belanja langsung, penduduk, regresi bergandaABSTRACT    The economy experienced a period of negative growth means economic activity in this period has decreased. Bitung-year period 2004-2014 economic growth fluctuations. These fluctuations can be influenced by private investment, direct spending, and population Economic growth is one measure of the success of economic development in an area. Economic growth reflects economic activity. Economic growth can be positive and can also be negative. If the economy experienced a period of positive growth means economic activity during the period has increased. Whereas if  years 2004-2014 are sourced from the Central Statistics Agency of North Sulawesi Province and Bitung. The analytical method used is an econometric model double-log regression (log-log) with Ordinary Least Square (OLS). This study aims to determine whether the development of private investment, direct spending, and population affect the economic growth of the city of Bitung. The data used using time series data period.    The results of the regression model of economic growth with the regression equation is LPDRB = - LINV 4.445 + 0.036 + 0.049 + 2.229 LPOP LBL. From these results show the development of private investment, direct expenditure and population positive and significant impact on economic growth of Bitung.Keywords: Economic growth, direct spending, population, regression.

An Efficient Method for Forecasting Using Fuzzy Time Series

2017 ◽  
pp. 287-304 ◽  
Author(s):  
Pritpal Singh
Keyword(s):  
Time Series ◽  
Time Series Data ◽  
Weather Forecasting ◽  
Small Error ◽  
Fuzzy Time Series ◽  
Series Data ◽  
Data Sets ◽  
Proposed Model ◽  
Temperature Forecasting ◽  
The University

Forecasting using fuzzy time series has been applied in several areas including forecasting university enrollments, sales, road accidents, financial forecasting, weather forecasting, etc. Recently, many researchers have paid attention to apply fuzzy time series in time series forecasting problems. In this paper, we present a new model to forecast the enrollments in the University of Alabama and the daily average temperature in Taipei, based on one-factor fuzzy time series. In this model, a new frequency based clustering technique is employed for partitioning the time series data sets into different intervals. For defuzzification function, two new principles are also incorporated in this model. In case of enrollments as well daily temperature forecasting, proposed model exhibits very small error rate.

Estimation of Most Probable Maximum From Short-Duration or Undersampled Time-Series Data

2015 ◽  
Author(s):  
Puneet Agarwal ◽  
William Walker ◽  
Kenneth Bhalla
Keyword(s):  
Time Series ◽  
Gaussian Process ◽  
Short Duration ◽  
Time Series Data ◽  
Extreme Value ◽  
Series Data ◽  
Sampled Data ◽  
Zero Crossing ◽  
Short Time ◽  
Undersampled Data

The most probable maximum (MPM) is the extreme value statistic commonly used in the offshore industry. The extreme value of vessel motions, structural response, and environment are often expressed using the MPM. For a Gaussian process, the MPM is a function of the root-mean square and the zero-crossing rate of the process. Accurate estimates of the MPM may be obtained in frequency domain from spectral moments of the known power spectral density. If the MPM is to be estimated from the time-series of a random process, either from measurements or from simulations, the time series data should be of long enough duration, sampled at an adequate rate, and have an ensemble of multiple realizations. This is not the case when measured data is recorded for an insufficient duration, or one wants to make decisions (requiring an estimate of the MPM) in real-time based on observing the data only for a short duration. Sometimes, the instrumentation system may not be properly designed to measure the dynamic vessel motions with a fine sampling rate, or it may be a legacy instrumentation system. The question then becomes whether the short-duration and/or the undersampled data is useful at all, or if some useful information (i.e., an estimate of MPM) can be extracted, and if yes, what is the accuracy and uncertainty of such estimates. In this paper, a procedure for estimation of the MPM from the short-time maxima, i.e., the maximum value from a time series of short duration (say, 10 or 30 minutes), is presented. For this purpose pitch data is simulated from the vessel RAOs (response amplitude operators). Factors to convert the short-time maxima to the MPM are computed for various non-exceedance levels. It is shown that the factors estimated from simulation can also be obtained from the theory of extremes of a Gaussian process. Afterwards, estimation of the MPM from the short-time maxima is explored for an undersampled process; however, undersampled data must not be used and only the adequately sampled data should be utilized. It is found that the undersampled data can be somewhat useful and factors to convert the short-time maxima to the MPM can be derived for an associated non-exceedance level. However, compared to the adequately sampled data, the factors for the undersampled data are less useful since they depend on more variables and have more uncertainty. While the vessel pitch data was the focus of this paper, the results and conclusions are valid for any adequately sampled narrow-banded Gaussian process.

Visualizing State of Time-Series Data by Supervised Gaussian Process Dynamical Models

2015 ◽  
Vol 19 (5) ◽  
pp. 688-696
Author(s):  
Nobuhiko Yamaguchi ◽  
Keyword(s):  
Time Series ◽  
Gaussian Process ◽  
Time Series Data ◽  
Series Data ◽  
Nonlinear Dimensionality Reduction ◽  
Probabilistic Representation ◽  
Dynamical Models ◽  
Motion Data ◽  
Dimensionality Reduction Technique ◽  
Representation Of Time

Gaussian Process Dynamical Models (GPDMs) constitute a nonlinear dimensionality reduction technique that provides a probabilistic representation of time series data in terms of Gaussian process priors. In this paper, we report a method based on GPDMs to visualize the states of time-series data. Conventional GPDMs are unsupervised, and therefore, even when the labels of data are available, it is not possible to use this information. To overcome the problem, we propose a supervised GPDM (S-GPDM) that utilizes both the data and their corresponding labels. We demonstrate experimentally that the S-GPDM can locate related motion data closer together than conventional GPDMs.

scholarly journals Time-Series Causality with Missing Data

2021 ◽  
Vol 6 (1) ◽  
pp. 1-4
Author(s):  
Bo Yuan Chang ◽  
Mohamed A. Naiel ◽  
Steven Wardell ◽  
Stan Kleinikkink ◽  
John S. Zelek
Keyword(s):  
Time Series ◽  
Missing Data ◽  
Missing Values ◽  
Time Series Data ◽  
Multivariate Time Series ◽  
Gaussian Process Regression ◽  
Series Data ◽  
Causal Relationships ◽  
Sampled Data ◽  
The Past

Over the past years, researchers have proposed various methods to discover causal relationships among time-series data as well as algorithms to fill in missing entries in time-series data. Little to no work has been done in combining the two strategies for the purpose of learning causal relationships using unevenly sampled multivariate time-series data. In this paper, we examine how the causal parameters learnt from unevenly sampled data (with missing entries) deviates from the parameters learnt using the evenly sampled data (without missing entries). However, to obtain the causal relationship from a given time-series requires evenly sampled data, which suggests filling the missing data values before obtaining the causal parameters. Therefore, the proposed method is based on applying a Gaussian Process Regression (GPR) model for missing data recovery, followed by several pairwise Granger causality equations in Vector Autoregssive form to fit the recovered data and obtain the causal parameters. Experimental results show that the causal parameters generated by using GPR data filling offers much lower RMSE than the dummy model (fill with last seen entry) under all missing values percentage, suggesting that GPR data filling can better preserve the causal relationships when compared with dummy data filling, thus should be considered when dealing with unevenly sampled time-series causality learning.

scholarly journals Relation Inference among Sensor Time Series in Smart Buildings with Metric Learning

2020 ◽  
Vol 34 (04) ◽  
pp. 4683-4690 ◽  
Author(s):  
Shuheng Li ◽  
Dezhi Hong ◽  
Hongning Wang
Keyword(s):  
Time Series ◽  
Demand Response ◽  
Real World ◽  
Time Series Data ◽  
Metric Learning ◽  
Fault Detection And Diagnosis ◽  
Series Data ◽  
Smart Building ◽  
Deep Metric Learning ◽  
Detection And Diagnosis

Smart Building Technologies hold promise for better livability for residents and lower energy footprints. Yet, the rollout of these technologies, from demand response controls to fault detection and diagnosis, significantly lags behind and is impeded by the current practice of manual identification of sensing point relationships, e.g., how equipment is connected or which sensors are co-located in the same space. This manual process is still error-prone, albeit costly and laborious.We study relation inference among sensor time series. Our key insight is that, as equipment is connected or sensors co-locate in the same physical environment, they are affected by the same real-world events, e.g., a fan turning on or a person entering the room, thus exhibiting correlated changes in their time series data. To this end, we develop a deep metric learning solution that first converts the primitive sensor time series to the frequency domain, and then optimizes a representation of sensors that encodes their relations. Built upon the learned representation, our solution pinpoints the relationships among sensors via solving a combinatorial optimization problem. Extensive experiments on real-world buildings demonstrate the effectiveness of our solution.

Terrorism and Wine Tourism: The Case of Museum Attendance

2018 ◽  
Vol 13 (4) ◽  
pp. 375-383 ◽  
Author(s):  
Olivier Gergaud ◽  
Florine Livat ◽  
Haiyan Song
Keyword(s):  
Time Series ◽  
Time Series Data ◽  
Terrorist Attacks ◽  
Series Data ◽  
Regression Estimation ◽  
Sample Period ◽  
Wine Tourism ◽  
The Impact ◽  
The City ◽  
Jel Classifications

AbstractIn this article, we use attendance data from La Cité du Vin, a wine museum in the city of Bordeaux, to assess the impact of the recent wave of terror that affected France on wine tourism. We use recent count regression estimation techniques suited for time series data to build a prediction model of the demand for attendance at this museum. We conclude that the institution lost about 5,000 visitors over 426 days, during which 14 successive terrorist attacks took place. This corresponds to almost 1% of the total number of visitors in the sample period. (JEL Classifications: L83, Z30)

Sensors to Events: Semantic Modeling and Recognition of Events from Data Streams

2016 ◽  
Vol 10 (04) ◽  
pp. 461-501 ◽  
Author(s):  
Om Prasad Patri ◽  
Anand V. Panangadan ◽  
Vikrambhai S. Sorathia ◽  
Viktor K. Prasanna
Keyword(s):  
Time Series ◽  
Real World ◽  
Data Streams ◽  
Time Series Data ◽  
Semantic Representation ◽  
Expressive Power ◽  
Sensor Data ◽  
Series Data ◽  
Formal Approach ◽  
Semantic Computing

Detecting and responding to real-world events is an integral part of any enterprise or organization, but Semantic Computing has been largely underutilized for complex event processing (CEP) applications. A primary reason for this gap is the difference in the level of abstraction between the high-level semantic models for events and the low-level raw data values received from sensor data streams. In this work, we investigate the need for Semantic Computing in various aspects of CEP, and intend to bridge this gap by utilizing recent advances in time series analytics and machine learning. We build upon the Process-oriented Event Model, which provides a formal approach to model real-world objects and events, and specifies the process of moving from sensors to events. We extend this model to facilitate Semantic Computing and time series data mining directly over the sensor data, which provides the advantage of automatically learning the required background knowledge without domain expertise. We illustrate the expressive power of our model in case studies from diverse applications, with particular emphasis on non-intrusive load monitoring in smart energy grids. We also demonstrate that this powerful semantic representation is still highly accurate and performs at par with existing approaches for event detection and classification.

Sign in / Sign up

Export Citation Format

Share Document