def write_file(suffix: str, recoverable: bool = True, readable: bool = False):
# set tree name and get data.
df_train, df_val, df_test = dt_util.init_dt_data(suffix)
# make the tree.
tree = DecisionTree(data=df_train + df_val,
max_depth=5,
min_node_size=50,
var_types=dt_util.var_types)
root_node = tree.split(node=tree.root_node)
# display tree to console.
dtns = NodeStorage(root_node=root_node,
fname=dt_util.get_filename(suffix),
header=dt_util.header,
var_types=dt_util.var_types)
dtns.print_tree_preorder(node=root_node)
if recoverable:
# write tree to file (recoverable).
dtns.tree_to_file(root_node=root_node)
if readable:
# write tree to file (readable).
test_pred = tree.predict_list(examples=df_test, root_node=root_node)
acc = dt_util.accuracy(dt_util.get_labels(df_test), test_pred)
dtns.tree_to_file_readable(root_node=root_node, acc=acc)
def read_file(fname: str,
write_readable: bool = False,
pos_data: List[str] = None) -> Node:
# read recoverable file and return the root node.
dtns = NodeStorage(fname=fname,
header=dt_util.get_header(),
var_types=dt_util.var_types)
root_node = dtns.file_to_tree()
# print tree to console to make sure it worked.
dtns.print_tree_preorder(root_node)
# make the tree object so we can use it for predictions.
tree = DecisionTree(root_node=root_node)
if write_readable:
# write tree to file (readable). can't get accuracy without the data.
# use the filename to figure out which testing data to use for pred/acc.
for suffix in pos_data:
if suffix in fname:
df_test = dt_util.init_dt_data(suffix)[2]
test_pred = tree.predict_list(examples=df_test,
root_node=root_node)
acc = dt_util.accuracy(dt_util.get_labels(df_test), test_pred)
break
dtns.tree_to_file_readable(root_node=root_node, acc=acc)
return tree
# prompt a guess, and record accuracy.
import dt_util
import sys
import random
# prompt the user for which testing set to use
data_options = ["full_cat", "seg_cat", "full", "seg", "big_cat"]
if len(sys.argv) < 2 or str(sys.argv[1]) not in data_options:
print("Expecting argument in ", data_options)
exit()
suffix = str(sys.argv[1])
# get the testing data and print the header
df_test = dt_util.init_dt_data(str(suffix))[2]
#print("\nVariables are ", dt_util.header[0:-1])
# store labels and predictions
y, p = [], []
for i in range(len(df_test)):
# generate prediction at random
guess = random.randint(0, 1)
# add it and the label to their lists
y.append(int(df_test[i][-1]))
p.append(guess)
# check the accuracy and give the number of items tested
print("Accuracy is " + str(dt_util.accuracy(y, p)))
print("over " + str(len(p)) + " predictions.")
# specify necessary arguments passed from command line
print("Expecting suffix argument: full_cat, seg_cat, full, or seg.")
suffix = str(sys.argv[1])
# set tree name and get data
df_train, df_val, df_test = dt_util.init_dt_data(suffix)
# initialize the tree with the training data.
# now that we have finalized the tree, train with both training & validation data
tree = DecisionTree(data=df_train+df_val, max_depth=5, min_node_size=50, var_types=dt_util.var_types)
print("tree initialization finished")
root_node = tree.split(node=tree.root_node)
print("tree split finished")
# display and store the tree
dtns = NodeStorage(root_node=root_node, fname=dt_util.get_filename(suffix), header=dt_util.get_header(), var_types=dt_util.var_types)
print("dtns initialization finished")
print("\nPrinting tree preorder")
dtns.print_tree_preorder(node=root_node)
#print("Printing tree inorder")
#dtns.print_tree_inorder(node=root_node)
print("\nComputing the accuracy")
test_labels = dt_util.get_labels(df_test)
test_pred = tree.predict_list(examples=df_test,root_node=root_node)
acc = dt_util.accuracy(test_labels,test_pred)
print("Accuracy is " + str(acc) + "\n")
print("attempting to write to file")
dtns.tree_to_file_readable(root_node=root_node, acc=acc)
print("finished writing tree to file")