This is Richard Warnung's Typepad Profile.

Join Typepad and start following Richard Warnung's activity

Richard Warnung

Recent Activity

Just as a follow up. I was able to apply the code provided. I just did not use the random draws as starting values but rather let the optimizer choose random initial points. I do something along the following lines. Not using the initial grid but init points only. This works perfect and really improved my tuning. Thank you so much for sharing! Best wishes, Richard
ba_search <- BayesianOptimization(fit_bayes,
bounds = bounds,
#init_grid_dt = initial_grid,
init_points = 10,
n_iter = 30,
acq = "ucb",
kappa = 1,
eps = 0.0,
verbose = TRUE)

Bayesian Optimization of Machine Learning Models

by Max Kuhn: Director, Nonclinical Statistics, Pfizer Many predictive and machine learning models have structural or tuning parameters that cannot be directly estimated from the data. For example, when using K-nearest neighbor model, there is no analytical estimator for K (the number of neighbo...

Hi, thanks for this extremely interesting post. I copied all the code that you have and it all worked to a certain point. Just after calling ba_search <- BayesianOptimization(...) I got the following error:
Error in Pred_list[[i]] <- This_Score_Pred$Pred :
attempt to select less than one element in integerOneIndex
Thus it worked for the first round and crashed in the second. Do you have the code clean somewhere? I am using the most recent versions of caret and rBayesianOptimization. Thank your for any comment.

Bayesian Optimization of Machine Learning Models

by Max Kuhn: Director, Nonclinical Statistics, Pfizer Many predictive and machine learning models have structural or tuning parameters that cannot be directly estimated from the data. For example, when using K-nearest neighbor model, there is no analytical estimator for K (the number of neighbo...

Richard Warnung is now following The Typepad Team

Jun 15, 2016

Subscribe to Richard Warnungâ€™s Recent Activity