Feature Exploration for Prediction of Potential Tuna Fishing Zones

Abstract

Prediction for potential fishing zone is one of the important activities concerning for the tuna fishing exploration, conservation and management. Accurate prediction will give more efficient in fishing activities. One of the way to predict is the classification techniques. Currently, as the state of the art, most of the methods utilize the chlorophyll and SST features. However, there are still other parameters that can be utilized. In this paper, the other parameters are then observed: ocean currents and salinity feature. First the results shows that, taking a part of ocean currents together with the chlorophyll and SST feature combination gives the improvement on the prediction. On other hand, this ocean currents feature is then substituted with the salinity, and the result shows that the combination between salinity, chlorophyll, and SST also increases the result. Finally, the ocean current and salinity parameters are combined together with chlorophyll and SST parameters and the result was surprising. It is found that the last feature combination which includes Chlorophyll, SST, Ocean current and salinity gives the highest result in classification (in Naï ve Bayes reaches 69.03%, Decision Tree reaches 82.32% and SVM reaches 68.30% of accuracy) compared to the "baseline" feature combination including only Chlorophyll and SST (in Naï ve Bayes reaches 57.44%, Decision Tree reaches 58.91% and SVM reaches 56.74% of accuracy). Therefore it is suggested that the proposed feature can be harnessed for the better prediction of potential fishing zone.

Spatial and temporal clustering analysis on chlorophyll-a data distribution

“…As noted previously, the oceanographic parameters that are often used in studies of the determination of potential fishing zones are sea surface temperature and chlorophyll-a. According to the results of such studies, chlorophyll-a is the most influential in determining potential fishing zone [12], [13]. This proves that data on the dis-tribution of chlorophyll-a are needed to determine potential areas of fishing.…”

Data Mining Analysis with Association Rules Method to Determine the Result of Fish Catch using FP-Growth Algorithm

“…But the classification results are still classified as fair classification category. In 2015 Fitrianah et al [18] conducted a potential zone study for tuna fishing using classification techniques, it was found that the last combination of features that include Chlorophyll, SST, the current Ocean and salinity gave the highest yield in classification (in Naïve Bayes reaching 69 , 03%, Decision Tree reached 82.32% and SVM achieved 68.30% accuracy) compared to "base" combination features including only Chlorophyll and SST (in Naïve Bayes reached 57.44%, Decision Tree reached 58.91% and SVM achieved 56.74% accuracy).…”

Oceanographic features selection to predict the tuna potential fishing zones using SFFS method

“…In the previous studies [5], determining the related features has been done by testing several combination of oceanographic feature sets. Then, the feature set that yields a high prediction accuracy has been chosen.…”

“…The last combination (f 4) gives the best result, which is 82.32% using Decision Tree classification. In addition to discover the best relevant feature to the Albacore tuna PFZ [5], in this research we also wish to discover the best relevant feature to other tuna species such as Bigeye tuna.…”

“…To be able to maintain the tuna fishing activities, many studies are proposed to assist the fishermen to find the tuna potential fishing zones (PFZ). Starting from the fishing management study [2], the prediction of tuna potential fishing zone (PFZ) [3], [4] and also the study of oceanographic parameter that affect the prediction of tuna PFZ [5].…”