def build_tree(part, scoref=entropy, beta=0):
if(len(part) == 0): return decisionnode()
best_gain = 0
best_criteria = None
best_sets = None
columns = len(part[0]) -1
for elem in part:
for i in range(columns):
try:
(set1, set2) = divideset(part, i, float(elem[i]))
except ValueError:
(set1, set2) = divideset(part, i, elem[i])
total = len(part)
pr = len(set1)/ float(total)
pl = len(set2)/float(total)
gain = scoref(part) - pr * scoref(set1) - pl * scoref(set2)
if(gain > best_gain):
best_gain = gain
best_criteria = (i, elem[i])
best_sets = (set1, set2)
if(best_gain > beta):
tree_r = build_tree(best_sets[0], scoref, beta)
tree_l = build_tree(best_sets[1], scoref, beta)
try:
return decisionnode(best_criteria[0], float(best_criteria[1]),tb=tree_r, fb=tree_l, gain=best_gain)
except ValueError:
return decisionnode(best_criteria[0], best_criteria[1],tb=tree_r, fb=tree_l, gain=best_gain)
else:
return decisionnode(results=unique_counts(part), gain=best_gain)
def buildtree_iter(part, scoref=entropy, beta=0):
if len(part)==0: return decisionnode()
node_list = []
sets_list = [[part, None, None]]
while sets_list:
best_criteria = None
best_sets = None
best_gain = 0
current_node = sets_list.pop(0)
data_set = current_node[0]
father = current_node[1]
side = current_node[2]
update_best_gain = False
for row in range(len(data_set)):
for column in range(len(data_set[row])-1):
try:
set1,set2=divideset(data_set,column,float(data_set[row][column]))
except ValueError:
set1,set2=divideset(data_set,column,data_set[row][column])
total = len(data_set)
pr = len(set1)/ float(total)
pl = len(set2)/float(total)
current_gain = scoref(data_set) - pr * scoref(set1) - pl * scoref(set2)
if best_gain < current_gain:
best_gain = current_gain
best_sets = (set1,set2)
best_criteria = (column, data_set[row][column])
update_best_gain=True
if best_gain > beta:
try:
node = decisionnode(col=best_criteria[0], value=float(best_criteria[1]), gain=best_gain)
except ValueError:
node = decisionnode(col=best_criteria[0], value=best_criteria[1], gain=best_gain)
node_list.append([node, father, side])
sets_list.append([best_sets[0], node, True])
sets_list.append([best_sets[1], node ,False])
else:
node = decisionnode(results=unique_counts(data_set), gain=best_gain)
node_list.append([node,father,side])
for node1 in node_list:
for node2 in node_list:
if node1[0]==node2[1]:
if node2[2]==True:
node1[0].tb=node2[0]
else:
node1[0].fb=node2[0]
return node_list[0][0]
def buildtree(rows, scoref=entropy):
if len(rows) == 0: return decisionnode()
current_score = scoref(rows)
#print 'entropy score =' + str(current_score)
# Set up some variables to track the best criteria
best_gain = 0.0
best_criteria = None
best_sets = None
column_count = len(rows[0]) - 1
for col in range(0, column_count):
# Generate the list of different values in
# this column
column_values = {}
for row in rows:
column_values[row[col]] = 1
# Now try dividing the rows up for each value
# in this column
for value in column_values.keys():
(set1, set2) = divideset(rows, col, value)
# Information gain
p = float(len(set1)) / len(rows)
gain = current_score - p * scoref(set1) - (1 - p) * scoref(set2)
if gain > best_gain and len(set1) > 0 and len(set2) > 0:
best_gain = gain
best_criteria = (col, value)
best_sets = (set1, set2)
# Create the sub branches
if best_gain > 0:
trueBranch = buildtree(best_sets[0])
falseBranch = buildtree(best_sets[1])
return decisionnode(col=best_criteria[0],
value=best_criteria[1],
tb=trueBranch,
fb=falseBranch)
else:
return decisionnode(results=uniquecounts(rows))
def buildtree(rows,scoref=entropy):
if len(rows)==0: return decisionnode()
current_score=scoref(rows)
#print 'entropy score =' + str(current_score)
# Set up some variables to track the best criteria
best_gain=0.0
best_criteria=None
best_sets=None
column_count=len(rows[0])-1
for col in range(0,column_count):
# Generate the list of different values in
# this column
column_values={}
for row in rows:
column_values[row[col]]=1
# Now try dividing the rows up for each value
# in this column
for value in column_values.keys():
(set1,set2)=divideset(rows,col,value)
# Information gain
p=float(len(set1))/len(rows)
gain=current_score-p*scoref(set1)-(1-p)*scoref(set2)
if gain>best_gain and len(set1)>0 and len(set2)>0:
best_gain=gain
best_criteria=(col,value)
best_sets=(set1,set2)
# Create the sub branches
if best_gain>0:
trueBranch=buildtree(best_sets[0])
falseBranch=buildtree(best_sets[1])
return decisionnode(col=best_criteria[0],value=best_criteria[1],
tb=trueBranch,fb=falseBranch)
else:
return decisionnode(results=uniquecounts(rows))
def buildrandomtree(rows,
kcandidates,
nmin,
pickcandidate=pick_candidate_gini):
rows = rows[:]
if len(rows) == 0: return decisionnode()
candidates = []
column_count = len(rows[0]) - 1
#print "number of columns = " + str(column_count)
#pick k random candidates
#candidate = (column_index,value)
for i in range(0, kcandidates):
random_index = random.randint(0, column_count - 1)
#get all unique values for a specific feature (column)
column_values = {}
for row in rows:
column_values[row[random_index]] = 1
#get a cutting point
cutting_point = get_cutting_point(column_values)
#print "rand feature index ="+str(random_index)+ "\n cutting point="+str(cutting_point)
#add to list of candidates
candidates.append((random_index, cutting_point))
#print candidates
#choose a candidate based on function given
chosen_candidate = pickcandidate(candidates, rows)
#print chosen_candidate
col = chosen_candidate[0]
value = chosen_candidate[1]
#split set based on the feature and value
(set1, set2) = divideset(rows, col, value)
#set1 = truebranch
trueBranch = None
#set2 = falsebranch
falseBranch = None
#print "item in leaf1 = " +str(len(set1))
#check if set1 has the min size
if (len(set1) <= nmin):
#do voting on the elements of of set1
#set and answer for this true branch
voting_result = get_voting_result(set1)
trueBranch = decisionnode(results=voting_result)
else:
#it means we need to grow this
trueBranch = buildrandomtree(set1, kcandidates, nmin, pickcandidate)
#print "item in leaf2 = " +str(len(set2))
#check if set2 has the min size
if (len(set2) <= nmin):
#do voting on the elements of of set2
#set and answer for this true branch
voting_result = get_voting_result(set2)
#uniquecounts
falseBranch = decisionnode(results=voting_result)
else:
#it means we need to grow this
falseBranch = buildrandomtree(set2, kcandidates, nmin, pickcandidate)
return decisionnode(col=col, value=value, tb=trueBranch, fb=falseBranch)
def buildrandomtree(rows,kcandidates,nmin,pickcandidate=pick_candidate_gini):
rows = rows[:]
if len(rows)==0: return decisionnode()
candidates = []
column_count=len(rows[0])-1
#print "number of columns = " + str(column_count)
#pick k random candidates
#candidate = (column_index,value)
for i in range(0,kcandidates):
random_index = random.randint(0,column_count-1)
#get all unique values for a specific feature (column)
column_values={}
for row in rows:
column_values[row[random_index]]=1
#get a cutting point
cutting_point = get_cutting_point(column_values)
#print "rand feature index ="+str(random_index)+ "\n cutting point="+str(cutting_point)
#add to list of candidates
candidates.append((random_index,cutting_point))
#print candidates
#choose a candidate based on function given
chosen_candidate = pickcandidate(candidates,rows)
#print chosen_candidate
col = chosen_candidate[0]
value = chosen_candidate[1]
#split set based on the feature and value
(set1,set2)=divideset(rows,col,value)
#set1 = truebranch
trueBranch = None
#set2 = falsebranch
falseBranch = None
#print "item in leaf1 = " +str(len(set1))
#check if set1 has the min size
if(len(set1)<=nmin):
#do voting on the elements of of set1
#set and answer for this true branch
voting_result = get_voting_result(set1)
trueBranch = decisionnode(results=voting_result)
else:
#it means we need to grow this
trueBranch = buildrandomtree(set1,kcandidates,nmin,pickcandidate)
#print "item in leaf2 = " +str(len(set2))
#check if set2 has the min size
if(len(set2)<=nmin):
#do voting on the elements of of set2
#set and answer for this true branch
voting_result = get_voting_result(set2)
#uniquecounts
falseBranch = decisionnode(results=voting_result)
else:
#it means we need to grow this
falseBranch = buildrandomtree(set2,kcandidates,nmin,pickcandidate)
return decisionnode(col=col,value=value,tb=trueBranch,fb=falseBranch)
