Why flotation kinetics models (phenomenological, mathematical and practical) have not been evaluated based on likelihood ratio test (LRT) and/or information criteria (IC) in order to consider the number of model parameters (complexity of models) and fitness together?
It has been seen for over decades that the sum square error (SSE) or goodness of fit was used as the only criterion for evaluation of the flotation modeling. However, literature shows that selection indices of a model are developed using likelihood ratio test (LRT) and information criteria (IC) (i.e., Bayesian information (BI), low of iterated logarithm (LI) and Akaike information (AI) indices). Why still minerals engineers don’t use these statistical tools for evaluation of molding?
I brought up the discussion shortly in my previous paper (10.1080/12269328.2017.1392900) but I'm going to present it in detail in IMPC 2018 as a proceeding paper entitled:
A NOVEL STATISTICAL INSIGHT TO MODELING OF FLOTATION KINETICS
FYI:
AIC formulated by the statistician Hirotugu Akaike; it was originally named "an information criterion". It was first announced by Akaike at a 1971 symposium, the proceedings of which were published in 1973.
As of October 2014, the 1974 paper had received more than 14000 citations in the Web of Science: making it the 73rd most-cited research paper of all time.
The other paper is the blow one which is cited more than 19000.
Gideon E. Schwarz, Estimating the Dimension of a Model, March 1978, The Annals of Statistics 6(2).
Dear Vinnett,
I appreciate your consideration, sharing your opinion and introducing the paper “Performance Evaluation of Basic Flotation Kinetic Models Using Advanced Statistical Techniques”.
Yes. In addition to evaluation of gof and complexity of models by means of different statistical techniques such as AIC, LRT, AICc and SC, your concern (estimation of model parameters) is very important, too. Unfortunately, the process of modeling in minerals engineering with respect to some topics (e.g., RTD or flotation kinetics) is on the way of only curve fitting (including some of my own studies) which should be changed. Therefore, in my opinion more critical studies are required to carry out on the flotation kinetics modeling by applying new insights to it.
As you know and even stated in the paper entitled above, there is not any single flotation kinetics model to be uniformly the best for all data sets. However, I do still believe that it is possible to define a UNIQUE flotation model to be valid for majority of ores.
Anyway, I’m always open to cooperation studies with this regard.
Thanks again!
Best wishes,
Ahmad Hassanzadeh
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers