Interpretation of Drop Size Predictions from a Random Forest Model Using Local Interpretable Model-Agnostic Explanations (LIME) in a Rotating Disc Contactor

Drop size is a crucial parameterfor the efficient design and operationof the rotating disc contactor (RDC) in liquid-liquid extraction.The current work focuses on providing local and global explanationsfor the prediction of the drop size in a rotating disc contactor (RDC).The Random Forest (RF) regression model is a robust machine learningalgorithm that can accurately capture complex relationships in thedata. However, the interpretability of the model is limited. In orderto address the issue of interpretability of the developed RF model,in the current work, we employed Local Interpretable Model-AgnosticExplanations (LIME) of the predictions of the RF model. This providesboth local and global views of the model and thereby helps one togain insights into the factors influencing predictions. We have providedlocal explanations depicting the impact of different attributes onthe prediction of the output for any given input example. We havealso obtained global feature importance, providing the top subsetof informative attributes. We have also developed local surrogatemodels incorporating second order attribute interactions. This hasprovided important information about the effect of interactions onthe drop size prediction. By augmenting the random forest model withLIME, it is possible to develop a more accurate and interpretablemodel for estimating the drop size in RDCs, ultimately leading toimproved performance and efficiency.