added RF surrogate

src/mobo_qm9.py

@@ -11,6 +11,7 @@
 from botorch.models.transforms.outcome import Standardize
 from gpytorch.mlls.sum_marginal_log_likelihood import SumMarginalLogLikelihood
 import pandas as pd
+from sklearn.ensemble import RandomForestRegressor
 
 from .data.cm_featurizer import get_coulomb_matrix
 from.data.soap_featurizer import get_soap
@@ -144,6 +145,17 @@ def get_surrogate_model(self, acq):
         fit_gpytorch_mll(mll)
 
         return model
+
+    def surrogate_RF(self,acq):
+
+        features = self.train_indices[acq]
+        targets =  self.correct_sign(self.targets[self.train_indices[acq]])
+        model = RandomForestRegressor(max_depth=10, min_samples_split=4, n_estimators=len(self.X_train), random_state=13)
+        model.fit(features,targets)
+        #preds = [tree.predict(self.X_test.values) for tree in rf.estimators_]
+        #mean_pred, sigma = np.mean(preds, axis=0), np.std(preds, axis=0)
+        #return mean_pred, sigma
+        return model 
 
     def correct_sign(self, Y):
         y_copy = Y.copy()