def main():
df = pd.read_csv('wine.csv')
#pp.dfExplore(df) #exploration function for paper analysis
pp.clean(df) #Normalizes data
#df.drop('resid_sugar', axis = 1, inplace=True)
#df.drop('free_sulf_d', axis = 1, inplace=True)
#df.drop('citric_acid', axis = 1, inplace=True)
#df.drop('density', axis = 1, inplace=True)
#df.drop('fx_acidity', axis = 1, inplace=True)
x = createX(df)
y = createY(df)
#Feature selection
temp = list(df)
print(temp)
fs.recursiveFeatureElimination(x, y)
fs.featureImportance(x, y)
#classifiers
results = clf.decisionTree(x, y)
confMatrixOutput('Decision Tree', results)
results = clf.supportVectorMachine(x, y)
confMatrixOutput('Support Vector Machine', results)
results = clf.ANN(x, y)
confMatrixOutput('Artificial Neural Network', results)
results = clf.randomForest(x, y)
confMatrixOutput('Random Forest', results)
results = clf.rulesBased(x, y)
confMatrixOutput('Rules Based Classifier', results)
pp.bayes(df) # discretizes all the data to low, medium, and high.
x = createX(df)
y = createY(df)
results = clf.bayesClassifier(x, y)
confMatrixOutput('Naive Bayesian Classifier', results)
def experiment6_1(train, test, f):
over_sampled_train = SMOTEOverSampling(train)
keep = decisionTreeFSelect(over_sampled_train)
keep = f(over_sampled_train[keep])
train = Standardization(over_sampled_train[keep])
test = Standardization(test[keep])
return randomForest(train, test)
def random_forest_depth_exp_SM_LV_ST_RF(train, test, max_depth):
over_sampled_train = SMOTEOverSampling(train)
keep = lowVarianceElimination(over_sampled_train, 0.8)
train = Standardization(over_sampled_train[keep])
test = Standardization(test[keep])
return randomForest(train, test, max_depth=max_depth)
def experiment17(train, test, f):
keep = f(train)
train = Standardization(train[keep])
test = Standardization(test[keep])
return randomForest(train, test)
def experiment18_1(train, test, f):
keep = decisionTreeFSelect(train)
keep = f(train[keep])
train = Standardization(train[keep])
test = Standardization(test[keep])
return randomForest(train, test)
def experiment18(train, test, f):
keep = univariateFSelect(train)
keep = f(train[keep])
train = Standardization(train[keep])
test = Standardization(test[keep])
return randomForest(train, test)
def experiment5(train, test, f):
over_sampled_train = SMOTEOverSampling(train)
keep = f(over_sampled_train)
train = Standardization(over_sampled_train[keep])
test = Standardization(test[keep])
return randomForest(train, test)
def experiment12_1(train, test, f):
over_sampled_train = SMOTEOverSampling(train)
keep = decisionTreeFSelect(over_sampled_train)
keep = f(over_sampled_train[keep])
return randomForest(over_sampled_train[keep], test[keep])
def experiment12(train, test, f):
over_sampled_train = SMOTEOverSampling(train)
keep = univariateFSelect(over_sampled_train)
keep = f(over_sampled_train[keep])
return randomForest(over_sampled_train[keep], test[keep])
def experiment11(train, test, f):
over_sampled_train = SMOTEOverSampling(train)
keep = f(over_sampled_train)
return randomForest(over_sampled_train[keep], test[keep])