def error(program: Dag) -> np.array:
errors = []
for ndx, (root, filenames) in enumerate(x_train):
y_true = y_train[ndx]
n_errors = len(y_true) + 1
if program is None:
errors += [penalty] * n_errors
else:
try:
y_pred = program.eval(root=root, filenames=filenames)
except Exception as e:
# print(e)
errors += [penalty] * n_errors
continue
# Compare the outputs element-wise and compute absolute errors.
for ndx, el_pred in enumerate(y_pred[:len(y_true)]):
el_true = y_true[ndx]
errors.append(
damerau_levenshtein_distance(el_true.path_str,
el_pred.path_str))
# If y_pred is shorter than y_true, fill in missing elements with penalties.
for _ in range(max(0, len(y_true) - len(y_pred))):
errors.append(penalty)
# Add the difference in size as an error.
errors.append(abs(len(y_true) - len(y_pred)))
return np.array(errors)
def error(program: Dag) -> np.array:
errors = []
for case in x_train:
y_true = target(case)
if program is None:
errors.append(penalty)
else:
try:
y_pred = program.eval(lst=case)
errors.append(abs(y_true - y_pred))
except Exception as e:
# print(e)
errors.append(penalty)
return np.array(errors)
def error(program: Dag) -> np.array:
errors = []
for i, dec in x_train:
if program is None:
errors.append(penalty)
else:
y_true = target(i, dec)
try:
y_pred = program.eval(i=i, dec=dec)
errors.append(abs(y_true - y_pred))
except Exception as e:
# print(e)
errors.append(penalty)
return np.array(errors)
def error(program: Dag) -> np.array:
errors = []
for dt1, dt2, y_true in X_train:
if program is None:
errors.append(penalty)
else:
try:
y_pred = program.eval(dt1=dt1, dt2=dt2)
errors.append(abs(y_pred - y_true))
except Exception as e:
# print(e)
errors.append(penalty)
continue
return np.array(errors)
def error(program: Dag) -> np.array:
errors = []
for x1, x2 in x_train:
if program is None:
errors.append(penalty)
errors.append(penalty)
else:
y_true = target(x1, x2)
try:
y_pred = program.eval(x1=float(x1), x2=float(x2))
errors.append(float(not isinstance(y_pred, float)))
errors.append(float((y_true - y_pred)**2))
except Exception as e:
# print(e)
errors.append(penalty)
errors.append(penalty)
return np.array(errors)
def error(program: Dag) -> np.array:
errors = []
for i in x_train:
if program is None:
errors.append(penalty)
errors.append(penalty)
else:
y_true = target(i)
try:
y_pred = program.eval(i=i)
errors.append(float(not isinstance(y_pred, bool)))
errors.append(float(not (bool(y_pred) == y_true)))
except Exception as e:
# print(e)
errors.append(penalty)
errors.append(penalty)
return np.array(errors)
def error(program: Dag) -> np.array:
errors = []
for lst, lower, upper, y_true in X_train:
n_errors = len(y_true) + 1
if program is None:
errors += [penalty] * n_errors
else:
try:
y_pred = program.eval(lst=lst, lower=lower, upper=upper)
except Exception as e:
# print(e)
errors += [penalty] * n_errors
continue
# Compare the outputs element-wise and compute absolute errors.
for ndx, el_pred in enumerate(y_pred[:len(y_true)]):
el_true = y_true[ndx]
errors.append(int(el_true != el_pred))
# If y_pred is shorter than y_true, fill in missing elements with penalties.
for _ in range(max(0, len(y_true) - len(y_pred))):
errors.append(penalty)
# Add the difference in size as an error.
errors.append(abs(len(y_true) - len(y_pred)))
return np.array(errors)