SONAR Signal Classification Using Logistic Regression

This kind of application is applied as the logistic regression to categorize the SONAR signals between Rocks and Mines and this paper aims at presenting the findings. It contains 208 samples with 60 features per sample with two classes, namely, Rocks represented as <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}$</tex> and Mines represented as M. Three folds of data splitting like 10:90, 20:80 and 30:70 are used in order to evaluate the proposed model. Accuracy, Precision, Recall, and Confusion Matrices are the important points of the evaluation. The results of the experiments support the fact that the model is applicable in all splits where the recall is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{8 8. 7 6}$</tex> percent in terms of accuracy <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{8 3. 7 3}$</tex> percent in terms of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2 0: 8 0}$</tex> split. As it can be concluded in the cases studied within the framework of this work, logistic regression model provides a decent starting point in case the SONAR signals are classified on with the relatively acceptable accuracy rates and understandable complexity. Future research will entail additional optimization of the selected model by choosing the relevant features, optimization of the model parameters and development of a single or combination of models, which will aim to enhance the model accuracy and reliability in the classification of different benign and malignant tumours.

Ҳали таржима қилинмаган