def train():
features = []
train = []
keys = {}
with open('train.csv', 'r') as dataset:
for x, i in enumerate(next(dataset).split(',')):
keys[i] = x # Initialize list of feature names
features = ([
next(dataset).strip('\n').replace(' ', '').split(',')
for x in range(35000)
])
#Way too messy
dataset.close()
for i in range(len(features)):
train.append(features[i].pop())
#Remove unwanted data
keep_list = [] #Only keep data points where room was booked
for i in range(len(features)):
if features[i][keys['is_booking']] == '1':
keep_list.append(i)
features = [features[i] for i in keep_list]
train = [train[i] for i in keep_list]
remove = ['date_time', 'srch_ci', 'srch_co', 'cnt']
features = zip(*features)
rem_list = []
for i in remove:
rem_list.append(keys[i])
for i in features:
if '' in i:
rem_list.append(features.index(i)) #Deal with nulls later
features = [features[i] for i in range(len(features)) if i not in rem_list]
features = zip(*features)
#Make CDT
CDT, feature_order, cluster_order = CA.make_CDT(features, train)
#Run MCA
u_p, lambda_s, u, chi_sq, psi_sq = CA.MCA(CDT)
#Project clusters onto principal axes
cluster_proj = CA.projections(CDT, u)
return cluster_proj, u, feature_order
