Exploration
reza Shahnavehsi; Farnusch Hajizadeh
Abstract
The present work is mainly about a method for illustrating the relation between the raw data in the same time; clustering is a key procedure to solve the problem of data division; also illustrating the connection among the elements of the research area simultaneously is important. Therefore, we propose ...
Read More
The present work is mainly about a method for illustrating the relation between the raw data in the same time; clustering is a key procedure to solve the problem of data division; also illustrating the connection among the elements of the research area simultaneously is important. Therefore, we propose a novel kind of clustering for data mining in the gravity field to reach the presenting connection among all elements in the same time. For this research work, 867 gravity surveying points were collected in the southern part of Iran (near diapir of Larestan) with a range of absolute gravity from 978579.672 to 978981.186. In this paper, clustering by self-organizing- maps, by utilizing scatter plot matrix is utilized for detecting the relation between the easting, northing, elevation, and absolute gravity simultaneously. In the proposed method, the relations between arrays, two by two, are defined, and like matrix, each raw and column has different i and j values, which represent elements of the studied area, instead of number; for example, array A23 is data division between i = 2 or raw two (in our case northing) and j = 3 or column, three (in our case elevation). In this algorithm, firstly, by using self-organizing maps, clustering is done, and this processing is generated to all arrays by scatter plot matrix, and in all arrays, three clusters are proposed; the result of this clustering shows that in arrays A12, A13, A14, A21, A23, A24, A31, A32, A41, A42, clustering is performed perfectly, and the relationship between the parameters of the studied area near Larestan salt, diaper, can be useful in notifying the properties of this salt diapir.
Exploration
Ahmadreza Erfan; Saeed Soltani Mohammad; Maliheh Abbaszadeh
Abstract
Machine learning (ML) has significantly transformed multiple disciplines, including mineral resource evaluation in mining engineering, by facilitating more accurate and efficient estimation methods. Ensemble methods, as a fundamental component of modern machine learning, have emerged as powerful ...
Read More
Machine learning (ML) has significantly transformed multiple disciplines, including mineral resource evaluation in mining engineering, by facilitating more accurate and efficient estimation methods. Ensemble methods, as a fundamental component of modern machine learning, have emerged as powerful tools that robust techniques that integrate multiple predictive models to improve performance beyond that of any individual learner. This study proposes a novel ensemble method for estimating ore grades by localizing the base learner weights in ensemble method. Ordinary kriging, inverse distance weighting, k-nearest neighbors, support vector regression, and artificial neural networks have been used as the base learners of the algorithm. In ML base learners, coordinates (easting, northing and elevation) of samples have been defined as input nodes and grade has been defined as target. The proposed method has been validated for predicting the copper grade (Cu%) in Darehzar porphyry deposit. The performance of proposed method has been by individual base learners and famous ensemble methods. This comparison shows that performance of proposed method is better than other ones. The findings highlight the necessity of adapting ensemble methods to address spatial variability in geological data, thereby establishing a robust framework for ore grade estimation.
Exploration
Mohammadjafar Mohammadzadeh; Majid Mahboubiaghdam; Moharram Jahangiri; Aynur Nasseri
Abstract
Most machine learning-monitored algorithms used to create mineral potential prediction maps require noise-free data to achieve high performance and reliable results. Unsupervised clustering methods are highly effective for uncovering a dataset’s hidden structures. Therefore, this study attempts ...
Read More
Most machine learning-monitored algorithms used to create mineral potential prediction maps require noise-free data to achieve high performance and reliable results. Unsupervised clustering methods are highly effective for uncovering a dataset’s hidden structures. Therefore, this study attempts a combination of supervised and unsupervised methods employing training and testing data to generate a highly accurate potential map of the Sonajil copper-gold deposit located in the NW of Iran. Here, a semi-supervised Bayesian algorithm is used to map the mineral landscape. Initially, ten raster layers of exploratory features are prepared. Then based on the copper concentration, 27 exploratory drilled boreholes are divided into four classes, C1 to C4, and from each class, two boreholes are selected, and 100-meter buffering is performed around these boreholes to extract 1113 training data based on the behavioral pattern of boreholes and surface samples. Subsequently, the existing data is clustered using the FCM method, and the total dataset and the clustering data are entered into the Bayesian algorithm to evaluate the accuracy of the Bayesian classifier method across five distinct clusters. The results show increased average accuracy when using clustered data instead of whole data for MPM mapping. Notably, the Bayesian semi-supervised algorithm achieved an impressive accuracy rate of 96% when cluster five data is excluded. To validate the Bayesian semi-supervised method, boreholes data that is not used in training were employed, which confirm the credibility of generated MPM. Overall results highlight the value of the Bayesian semi-supervised algorithm in improving the accuracy and reliability of mineral prospectivity mapping via the application of the FCM clustering method that efficiently organize the data, enabling the Bayesian algorithm to evaluate the accuracy of the Bayesian classifier method across different clusters and providing a successful optimal result in detecting blind ores in areas without exploratory boreholes and delineating more mineralization targets in the Sonajil and adjoining areas.
