In this paper we have presented a novel approach to predict dissolved oxygen in prawn ponds.It is necessary to maintain dissolved oxygen above a certain level in the ponds for expected growth and survival of the prawn...In this paper we have presented a novel approach to predict dissolved oxygen in prawn ponds.It is necessary to maintain dissolved oxygen above a certain level in the ponds for expected growth and survival of the prawns.An accurate prediction of dissolved oxygen can assist farmers to take necessary measures to maintain dissolved oxygen levels ideal for prawn growth.Existing approaches to dissolved oxygen prediction performs well on short term,however incurs high error on long term prediction.We propose a new approach where a group of predictors are developed where each model predicts a certain time stamps ahead.Each predictor is trained on sampled data so that it predicts a step ahead prediction only,however,the sampling process decides on the actual number of time stamp ahead prediction.Since step ahead predictor acts like a short term predictor,it incurs small error even at higher time stamp ahead prediction.Experimental results demonstrate that the proposed approach achieves significantly lower error on long term prediction compared to other existing approaches.展开更多
Shellfish farms are closed for harvest when microbial pollutants are present.Such pollutants are typically present in rainfall runoff from various land uses in catchments.Experts currently use a number of observable p...Shellfish farms are closed for harvest when microbial pollutants are present.Such pollutants are typically present in rainfall runoff from various land uses in catchments.Experts currently use a number of observable parameters(river flow,rainfall,salinity)as proxies to determine when to close farms.We have proposed using the short term historical rainfall data as a time-series prediction problem where we aim to predict the closure of shellfish farms based only on rainfall.Time-series event prediction consists of two steps:(i)feature extraction,and(ii)prediction.A number of data mining challenges exist for these scenarios:(i)which feature extraction method best captures the rainfall pattern over successive days that leads to opening or closure of the farms?,(ii)The farm closure events occur infrequently and this leads to a class imbalance problem;the question is what is the best way to deal with this problem?In this paper we have analysed and compared different combinations of balancing methods(under-sampling and over-sampling),feature extraction methods(cluster profile,curve fitting,Fourier Transform,Piecewise Aggregate Approximation,and Wavelet Transform)and learning algorithms(neural network,support vector machine,k-nearest neighbour,decision tree,and Bayesian Network)to predict closure events accurately considering the above data mining challenges.We have identified the best combination of techniques to accurately predict shellfish farm closure from rainfall,given the above data mining challenges.展开更多
文摘In this paper we have presented a novel approach to predict dissolved oxygen in prawn ponds.It is necessary to maintain dissolved oxygen above a certain level in the ponds for expected growth and survival of the prawns.An accurate prediction of dissolved oxygen can assist farmers to take necessary measures to maintain dissolved oxygen levels ideal for prawn growth.Existing approaches to dissolved oxygen prediction performs well on short term,however incurs high error on long term prediction.We propose a new approach where a group of predictors are developed where each model predicts a certain time stamps ahead.Each predictor is trained on sampled data so that it predicts a step ahead prediction only,however,the sampling process decides on the actual number of time stamp ahead prediction.Since step ahead predictor acts like a short term predictor,it incurs small error even at higher time stamp ahead prediction.Experimental results demonstrate that the proposed approach achieves significantly lower error on long term prediction compared to other existing approaches.
文摘Shellfish farms are closed for harvest when microbial pollutants are present.Such pollutants are typically present in rainfall runoff from various land uses in catchments.Experts currently use a number of observable parameters(river flow,rainfall,salinity)as proxies to determine when to close farms.We have proposed using the short term historical rainfall data as a time-series prediction problem where we aim to predict the closure of shellfish farms based only on rainfall.Time-series event prediction consists of two steps:(i)feature extraction,and(ii)prediction.A number of data mining challenges exist for these scenarios:(i)which feature extraction method best captures the rainfall pattern over successive days that leads to opening or closure of the farms?,(ii)The farm closure events occur infrequently and this leads to a class imbalance problem;the question is what is the best way to deal with this problem?In this paper we have analysed and compared different combinations of balancing methods(under-sampling and over-sampling),feature extraction methods(cluster profile,curve fitting,Fourier Transform,Piecewise Aggregate Approximation,and Wavelet Transform)and learning algorithms(neural network,support vector machine,k-nearest neighbour,decision tree,and Bayesian Network)to predict closure events accurately considering the above data mining challenges.We have identified the best combination of techniques to accurately predict shellfish farm closure from rainfall,given the above data mining challenges.