Hod Lipson, one of the creators of Eureqa, a symbolic regression (genetic programming) code package, ran some tests of different machine learning algorithms. He used seven data sets from the UC Irvine Machine Learning Repository. Results were presented in “How does Eureqa Compare to Other Machine Learning Methods?“:
‘How does Eureqa’s performance, in terms of predictive accuracy and simplicity, compare to other machine learning methods, such as Neural Networks, Support Vector Machines, Decision Trees, and simple Linear Regression?
To answer this question we did a simple comparison. We ran Eureqa on seven test-cases for which data is publically available, and compared performance to four standard machine learning methods. The implementations used were the WEKA codes, with settings optimized for best performance.‘
‘It appears that Eureqa’s use of symbolic regression produces models that are both more accurate and simpler than other machine learning methods, but what’s the catch?
There is no free lunch. Symbolic regression is substantially more computationaly intensive when compared to neural networks, SVMs and Linear regression.‘
Unfortunately, Lipson does not provide details regarding the parameters used for each algorithm. Thus, one cannot independently assess his conclusions. However, from my own experience and knowledge of the algorithms used, I believe that the broad conclusions contained in the graph above are accurate. I would also note that such conclusions are heavily dependent on the data sets used.