Pattern Sequence-based Forecasting Research Articles

Natural Time Series Parameters Forecasting: Validation of the Pattern-Sequence-Based Forecasting (PSF) Algorithm; A New Python Package

Climate change has contributed substantially to the weather and land characteristic phenomena. Accurate time series forecasting for climate and land parameters is highly essential in the modern era for climatologists. This paper provides a brief introduction to the algorithm and its implementation in Python. The pattern-sequence-based forecasting (PSF) algorithm aims to forecast future values of a univariate time series. The algorithm is divided into two major processes: the clustering of data and prediction. The clustering part includes the selection of an optimum value for the number of clusters and labeling the time series data. The prediction part consists of the selection of a window size and the prediction of future values with reference to past patterns. The package aims to ease the use and implementation of PSF for python users. It provides results similar to the PSF package available in R. Finally, the results of the proposed Python package are compared with results of the PSF and ARIMA methods in R. One of the issues with PSF is that the performance of forecasting result degrades if the time series has positive or negative trends. To overcome this problem difference pattern-sequence-based forecasting (DPSF) was proposed. The Python package also implements the DPSF method. In this method, the time series data are first differenced. Then, the PSF algorithm is applied to this differenced time series. Finally, the original and predicted values are restored by applying the reverse method of the differencing process. The proposed methodology is tested on several complex climate and land processes and its potential is evidenced.

Time series forecasting method based on frequent pattern mining

Pattern Sequence-based Forecasting (PSF) is an effective method for time series prediction. However, the accuracy of this method depends on the selection of parameters such as the length of the pattern sequence and the number of clusters. In diverse time series data sets, these parameters are often priori unknown. This paper innovatively introduces a pattern mining method before the PSF pattern clustering to guide the clustering process and realize the automation of initial parameter selection. Experimental results show that the method proposed in this paper effectively eliminates the uncertainty of PSF caused by the selection of initial parameters. Compared with the original model, it improves the efficiency while ensuring the advantage of prediction accuracy.

The Hybridization of Ensemble Empirical Mode Decomposition with Forecasting Models: Application of Short-Term Wind Speed and Power Modeling

In this research, two hybrid intelligent models are proposed for prediction accuracy enhancement for wind speed and power modeling. The established models are based on the hybridisation of Ensemble Empirical Mode Decomposition (EEMD) with a Pattern Sequence-based Forecasting (PSF) model and the integration of EEMD-PSF with Autoregressive Integrated Moving Average (ARIMA) model. In both models (i.e., EEMD-PSF and EEMD-PSF-ARIMA), the EEMD method is used to decompose the time-series into a set of sub-series and the forecasting of each sub-series is initiated by respective prediction models. In the EEMD-PSF model, all sub-series are predicted using the PSF model, whereas in the EEMD-PSF-ARIMA model, the sub-series with high and low frequencies are predicted using PSF and ARIMA, respectively. The selection of the PSF or ARIMA models for the prediction process is dependent on the time-series characteristics of the decomposed series obtained with the EEMD method. The proposed models are examined for predicting wind speed and wind power time-series at Maharashtra state, India. In case of short-term wind power time-series prediction, both proposed methods have shown at least 18.03 and 14.78 percentage improvement in forecast accuracy in terms of root mean square error (RMSE) as compared to contemporary methods considered in this study for direct and iterated strategies, respectively. Similarly, for wind speed data, those improvement observed to be 20.00 and 23.80 percentages, respectively. These attained prediction results evidenced the potential of the proposed models for the wind speed and wind power forecasting. The current proposed methodology is transformed into R package ‘decomposedPSF’ which is discussed in the Appendix.

Prediction Of Trends In Bus Travel Time Using Spatial Patterns

Studying patterns in traffic data is a basic analysis to understand the system. In this study, a large amount of bus travel data collected using vehicle tracking devices is analyzed for patterns. Travel time, in general, follows both spatial and temporal patterns. Spatial patterns are expected because travel times in particular sections on a roadway can be following similar patterns. For example, sections with a bus stop in it may show similar patterns due to stopping at the bus stops. The present study explores the use of data-driven approaches, primarily clustering, to identify the spatial patterns in bus travel times. Discrete Wavelet Transform (DWT) is used to extract trends from the travel time measurements. Two popular clustering algorithms - k-means and hierarchical clustering algorithms are used in this study to identify the spatial patterns and group sections with similar characteristics. Once the spatial patterns are obtained, the historic database is searched to identify similar cluster patterns and travel time trends are predicted using Pattern Sequence-based Forecasting (PSF) algorithm. The performance of the proposed algorithm for the prediction of travel time trends of trips occurring during peak and off-peak hours of a day was then compared based on prediction errors.

Soft computing applications in the field of industrial and environmental enterprises

Soft computing applications in the field of industrial and environmental enterprises

Multi-step forecasting for big data time series based on ensemble learning

This paper presents ensemble models for forecasting big data time series. An ensemble composed of three methods (decision tree, gradient boosted trees and random forest) is proposed due to the good results these methods have achieved in previous big data applications. The weights of the ensemble are computed by a weighted least square method. Two strategies related to the weight update are considered, leading to a static or dynamic ensemble model. The predictions for each ensemble member are obtained by dividing the forecasting problem into h forecasting sub-problems, one for each value of the prediction horizon. These sub-problems have been solved using machine learning algorithms from the big data engine Apache Spark, ensuring the scalability of our methodology. The performance of the proposed ensemble models is evaluated on Spanish electricity consumption data for 10 years measured with a 10-minute frequency. The results showed that both the dynamic and static ensembles performed well, outperforming the individual ensemble members they combine. The dynamic ensemble was the most accurate model achieving a MRE of 2%, which is a very promising result for the prediction of big time series. Proposed ensembles are also evaluated using solar power from Australia for two years measured with 30-min frequency. The results are successfully compared with Artificial Neural Network, Pattern Sequence-based Forecasting and Deep Learning, improving their results.

Evaluation of Clustering Algorithms for the Prediction of Trends in Bus Travel Time

Providing accurate and reliable travel time information to travellers is essential to improve the quality of public transit systems. With the availability of the latest technologies, it has become possible to collect a large amount of traffic data to analyze and understand these systems better. Traffic in India is characterized by lack of lane discipline and the presence of vehicles of varying static and dynamic characteristics, which makes prediction of bus travel time especially challenging. The aim of this study is to identify both a prediction algorithm that can handle high variability and suitable inputs or regressors to be used. Earlier studies performed offline manual grouping considering the patterns observed, which leads to limitations for automated field implementations. The present study explores the use of data-driven approaches, primarily clustering, to address the challenges for the prediction of bus travel time trends. Discrete wavelet transform (DWT) was used to extract trends from the travel time measurements. Three popular clustering algorithms— k-means, hierarchical, and self organizing maps (SOM)—were used to identify patterns. Travel time trends were then predicted by searching for similar cluster patterns within the historical database using pattern sequence-based forecasting (PSF). A comparison of the performance of these algorithms was carried out based on prediction errors. The clustering +prediction framework developed was also compared with the case when no clustering was done on the regressor dataset.

Analysis of differencing and decomposition preprocessing methods for wind speed prediction

Preprocessing methods improve prediction accuracy in a significant way. Generally, they stabilize data series mean and variance and remove irregularities. In this paper, two of these methods have been used and compared in the Pattern Sequence based Forecasting (PSF) algorithm. These preprocessing methods are based on differencing and decomposition principles. In the decomposition approach, the Ensemble Empirical Mode Decomposition-Pattern Sequence based Forecasting (EEMD-PSF) model is used. It decomposes the data into finite numbers of subseries and improves the performance of the PSF method. In the Difference Pattern Sequence based Forecasting (DPSF) method, the effects of trends, seasonality, and irregularity components are reduced to some extent and their consequences are tested on distinct datasets with different patterns. While comparing the effects of these preprocessing methods with the effect of the PSF method for sets of wind speed data collected in the autonomous region of Galicia, Spain, and National Renewable Energy Laboratory (NREL), USA, in terms of prediction accuracy, both methods have performed better than the contemporary methods including single PSF, ARIMA, and LSSVM methods. In terms of computational time consumption, the DPSF method has outperformed the results of the EEMD-PSF model. The simulations revealed that the hybridization of preprocessing and PSF methods has significantly outperformed other state-of-the-art methods for short term wind speed prediction.

Hybrid Leakage Management for Water Network Using PSF Algorithm and Soft Computing Techniques

Reduction of leakages in water distribution system (WDS) is one of the major concerns for water industries. This paper presents a hybrid leakage reduction model using pressure management technique, performed by optimizing water storage level in the tank, along with optimized control and localization of pressure reducing valve (PRV) in water distribution system. Pattern Sequence-based Forecasting (PSF) algorithm is used for prediction of flow rate (demand) from the tank for next 48 h, to calibrate the future desire water storage level in the tank. A mathematical tank and pump simulation algorithm is proposed for optimization of water storage level in the tank. A modified reference pressure algorithm is proposed for efficient localization of pressure reducing valve. Multiobjective genetic algorithm (NSGA-II) is used for finding out the optimized operational control setting of the pressure reducing valve for leakage minimization. The proposed algorithm leads to better leakage reduction of 20.81% in modified benchmark WDS, with a reduced number of the pressure reducing valves. Constraints such as maintaining lower hydraulic failure index (<0.01), emergency water storage, etc. is also considered. It can be concluded that the proposed hybrid leakage reduction technique provides efficient as well as cost-effective solution for leakage control.

Ensemble Prediction Model with Expert Selection for Electricity Price Forecasting

Forecasting of electricity prices is important in deregulated electricity markets for all of the stakeholders: energy wholesalers, traders, retailers and consumers. Electricity price forecasting is an inherently difficult problem due to its special characteristic of dynamicity and non-stationarity. In this paper, we present a robust price forecasting mechanism that shows resilience towards the aggregate demand response effect and provides highly accurate forecasted electricity prices to the stakeholders in a dynamic environment. We employ an ensemble prediction model in which a group of different algorithms participates in forecasting 1-h ahead the price for each hour of a day. We propose two different strategies, namely, the Fixed Weight Method (FWM) and the Varying Weight Method (VWM), for selecting each hour’s expert algorithm from the set of participating algorithms. In addition, we utilize a carefully engineered set of features selected from a pool of features extracted from the past electricity price data, weather data and calendar data. The proposed ensemble model offers better results than the Autoregressive Integrated Moving Average (ARIMA) method, the Pattern Sequence-based Forecasting (PSF) method and our previous work using Artificial Neural Networks (ANN) alone on the datasets for New York, Australian and Spanish electricity markets.

PSF: Introduction to R Package for Pattern Sequence Based Forecasting Algorithm

This paper discusses about an R package that implements the Pattern Sequence based Forecasting (PSF) algorithm, which was developed for univariate time series forecasting. This algorithm has been successfully applied to many different fields. The PSF algorithm consists of two major parts: clustering and prediction. The clustering part includes selection of the optimum number of clusters. It labels time series data with reference to such clusters. The prediction part includes functions like optimum window size selection for specific patterns and prediction of future values with reference to past pattern sequences. The PSF package consists of various functions to implement the PSF algorithm. It also contains a function which automates all other functions to obtain optimized prediction results. The aim of this package is to promote the PSF algorithm and to ease its implementation with minimum efforts. This paper describes all the functions in the PSF package with their syntax. It also provides a simple example of usage. Finally, the usefulness of this package is discussed by comparing it to auto.arima and ets, well-known time series forecasting functions available on CRAN repository.

Improved pattern sequence‐based forecasting method for electricity load

Recently, a pattern sequence‐based forecasting (PSF) algorithm was proposed for day‐ahead electricity time series. PSF consists of two steps: clustering and prediction. However, it has the following limitations: In the clustering step, it is computationally expensive to determine the optimal number of clusters with majority votes. In the prediction step, it is quite complex to search for the matched pattern sequence with the optimal window length, and averaging all the samples immediately after the matched sequence can increase the forecasting accuracy especially when the day under examination is a working day.In this paper, we propose a time‐series forecasting method for electricity load by addressing the limitations in PSF. The proposed method is evaluated on electricity load datasets, and the experimental results show that the proposed method can improve the forecasting accuracy of PSF. © 2014 Institute of Electrical Engineers of Japan. Published by John Wiley &amp; Sons, Inc.