# Split training data
train, test, train_lab, test_lab = train_test_split(Y,
emnist.labels,
stratify=emnist.labels,
test_size=0.3,
random_state=random_state)
#D = {"T":0, "t":1, "N":2, "n":3}
D = {'a': 0, 'b': 1, 'Cc': 2, 'd': 3, 'e': 4, 'A': 5, 'B': 6, 'D': 7, 'E': 8}
model = bs.BayesSoftMaxClassifier(train,
np.array([D[i] for i in train_lab]),
numClasses=9,
numSim=10000,
burnIn=5000,
candVar=.008,
paramVar=.06)
model.SamplePosterior()
# Evaluation
model.Predict(test)
err = np.mean(model.predictions != np.array([D[i] for i in test_lab]))
bal = balanced_accuracy_score(np.array([D[i] for i in test_lab]),
model.predictions)
print("Error in BayesSoftmax: " + str(err))
model.PlotParamTrace(-1)
model.PlotParamDist(0)
model.PlotPredictiveDistribution(0)
# Import libraries from sklearn.metrics import accuracy_score from sklearn.ensemble import RandomForestClassifier # define data, split to train & test, create model and fit data X = Variables Y = Classes Model = RandomForestClassifier(n_estimators=10, params).fit(X, Y) # Predict new classes Y_predict = Model.Predict(X_test) # Score model accuracy_score(Y_test, Y_predict)