Reprocessing of iron ore tailings (IOTs) and extracting recoverable valuable iron oxides will become increasingly financially attractive for mining companies and also may reduce environmental problems. Using databases built based on long term monitoring of units installed on plants to control the operational conditions to generate artificial intelligence models can decrease the cost of reprocessing operations Although some investigations have been focused on the reprocessing of IOTs, several challenges still remain which need to be addressed, especially for fine particles. SLon®, has developed a pulsating high gradient magnetic separator for the processing of fine iron oxides. However, there has been no systematic optimisation and variable assessments for SLon® operating variables to examine their effects on metallurgical responses (separation efficiency) on the industrial scale. This study addressed these drawbacks by linear (Pearson correlation) and non-linear (random forest) variable importance measurements (VIM) through an industrial SLon® installation.