Enhancing Fuzzy Associative Rule Mining Approaches for Improving Prediction Accuracy. Integration of Fuzzy Clustering, Apriori and Multiple Support Approaches to Develop an Associative Classification Rule Base
AuthorSowan, Bilal I.
SupervisorDahal, Keshav P.
Hossain, M. Alamgir
Fuzzy associative rule mining
Associative classification rule base
Decision support system
Minimizing prediction error
The University of Bradford theses are licenced under a Creative Commons Licence.
InstitutionUniversity of Bradford
DepartmentSchool of Computing, Informatics & Media
MetadataShow full item record
AbstractBuilding an accurate and reliable model for prediction for different application domains, is one of the most significant challenges in knowledge discovery and data mining. This thesis focuses on building and enhancing a generic predictive model for estimating a future value by extracting association rules (knowledge) from a quantitative database. This model is applied to several data sets obtained from different benchmark problems, and the results are evaluated through extensive experimental tests. The thesis presents an incremental development process for the prediction model with three stages. Firstly, a Knowledge Discovery (KD) model is proposed by integrating Fuzzy C-Means (FCM) with Apriori approach to extract Fuzzy Association Rules (FARs) from a database for building a Knowledge Base (KB) to predict a future value. The KD model has been tested with two road-traffic data sets. Secondly, the initial model has been further developed by including a diversification method in order to improve a reliable FARs to find out the best and representative rules. The resulting Diverse Fuzzy Rule Base (DFRB) maintains high quality and diverse FARs offering a more reliable and generic model. The model uses FCM to transform quantitative data into fuzzy ones, while a Multiple Support Apriori (MSapriori) algorithm is adapted to extract the FARs from fuzzy data. The correlation values for these FARs are calculated, and an efficient orientation for filtering FARs is performed as a post-processing method. The FARs diversity is maintained through the clustering of FARs, based on the concept of the sharing function technique used in multi-objectives optimization. The best and the most diverse FARs are obtained as the DFRB to utilise within the Fuzzy Inference System (FIS) for prediction. The third stage of development proposes a hybrid prediction model called Fuzzy Associative Classification Rule Mining (FACRM) model. This model integrates the ii improved Gustafson-Kessel (G-K) algorithm, the proposed Fuzzy Associative Classification Rules (FACR) algorithm and the proposed diversification method. The improved G-K algorithm transforms quantitative data into fuzzy data, while the FACR generate significant rules (Fuzzy Classification Association Rules (FCARs)) by employing the improved multiple support threshold, associative classification and vertical scanning format approaches. These FCARs are then filtered by calculating the correlation value and the distance between them. The advantage of the proposed FACRM model is to build a generalized prediction model, able to deal with different application domains. The validation of the FACRM model is conducted using different benchmark data sets from the University of California, Irvine (UCI) of machine learning and KEEL (Knowledge Extraction based on Evolutionary Learning) repositories, and the results of the proposed FACRM are also compared with other existing prediction models. The experimental results show that the error rate and generalization performance of the proposed model is better in the majority of data sets with respect to the commonly used models. A new method for feature selection entitled Weighting Feature Selection (WFS) is also proposed. The WFS method aims to improve the performance of FACRM model. The prediction performance is improved by minimizing the prediction error and reducing the number of generated rules. The prediction results of FACRM by employing WFS have been compared with that of FACRM and Stepwise Regression (SR) models for different data sets. The performance analysis and comparative study show that the proposed prediction model provides an effective approach that can be used within a decision support system.
Showing items related by title, author, creator and subject.
Rationalization of Racemate Resolution: Predicting Spontaneous Resolution through Crystal Structure Prediction.Kendrick, John; Gourlay, Matthew D.; Leusen, Frank J.J. (2009-07-14)Crystal structure prediction simulations are reported on 5-hydroxymethyl-2-oxazolidinone and 4-hydroxymethyl-2-oxazolidinone to establish the feasibility of predicting the spontaneous resolution of racemates of small organic molecules. It is assumed that spontaneous resolution occurs when the enantiomorph is more stable than the racemic solid. The starting point is a gas phase conformational search to locate all low-energy conformations. These conformations are used to predict the possible crystal structures of 5- and 4-hydroxymethyl-2-oxazolidinone. In both cases, the racemic crystal structure is predicted to have the lowest energy. The energy differences between the lowest-energy racemic solids and the lowest-energy enantiomorphs are 0.2 kcal mol-1 for 5-hydroxymethyl-2-oxazolidinone and 0.9 kcal mol-1 for 4-hydroxymethyl-2-oxazolidinone. In the case of 4-hydroxymethyl-2-oxazolidinone, where the racemic crystal is known to be more stable and the experimental crystal structures of both the racemate and the enantiomorph are available, the simulation results match the observed data. For 5-hydroxymethyl-2-oxazolidinone, where only enantiopure crystals are observed experimentally, the known experimental structure is found 1.6 kcal mol-1 above the lowest-energy predicted structure. This work shows that it is possible to predict whether the racemate of a small chiral molecule can be resolved spontaneously, although further advances in the accuracy of lattice energy calculations are required.
Automated Solar Activity Prediction: A hybrid computer platform using machine learning and solar imaging for automated prediction of solar flaresColak, Tufan; Qahwaji, Rami S.R. (AGU, 04/06/2009)The importance of real-time processing of solar data especially for space weather applications is increasing continuously. In this paper, we present an automated hybrid computer platform for the short-term prediction of significant solar flares using SOHO/Michelson Doppler Imager images. This platform is called the Automated Solar Activity Prediction tool (ASAP). This system integrates image processing and machine learning to deliver these predictions. A machine learning-based system is designed to analyze years of sunspot and flare data to create associations that can be represented using computer-based learning rules. An imaging-based real-time system that provides automated detection, grouping, and then classification of recent sunspots based on the McIntosh classification is also created and integrated within this system. The properties of the sunspot regions are extracted automatically by the imaging system and processed using the machine learning rules to generate the real-time predictions. Several performance measurement criteria are used and the results are provided in this paper. Also, quadratic score is used to compare the prediction results of ASAP with NOAA Space Weather Prediction Center (SWPC) between 1999 and 2002, and it is shown that ASAP generates more accurate predictions compared to SWPC.
Predicting spontaneous racemate resolution using recent developments in crystal structure predictionKendrick, John; Gourlay, Matthew D.; Neumann, M.A.; Leusen, Frank J.J. (2009)A hybrid molecular mechanics and quantum mechanics solid state DFT method is used to re-rank the stability of racemic and enantiopure crystal structures of four molecules; 4-hydroxymethyl-2-oxazolidinone, 5-hydroxymethyl-2-oxazolidinone, 2-(4-hydroxyphenyl)-2,5,5-trimethylpyrrolidine-1-oxy and 2-(3-hydroxyphenyl)-2,5,5-trimethylpyrrolidine-1-oxy. Previous work using a force field based method to predict these crystal structures indicated that the lattice energy may be a suitable criterion for predicting whether a chiral molecule will resolve spontaneously on crystallisation. However, in some cases, the method had predicted an unrealistically high lattice energy for the structure corresponding to the experimentally observed one. The Hybrid DFT method successfully predicts those molecules which resolve spontaneously and furthermore predicts satisfactory lattice energies for all experimentally observed structures. Based on a comparison of the predicted lattice energies from the two methods it is concluded that the force fields used were not sufficiently accurate to predict spontaneous resolution with any confidence. However, the Hybrid DFT method is shown to be sufficiently accurate for making such predictions.