def financial_explan_2_per_client(case_id):
form = FlaskForm(request.form)
form.case_id = case_id
form.categories = json.loads(session.get("categories", "{}"))
form.subdomain = Subdomain.query.get(session.get("subdomain", None))
ranges_for_feature = session.get("ranges_for_feature", {})
range_groups_ages = ranges_for_feature.get("Age", [])
all_case_values_query = CaseValue.query.filter(
CaseValue.user_data_id == session["user_data_id"])
all_case_values_query_for_patient = all_case_values_query.filter(
CaseValue.case_id == case_id)
form.table_heads = [
"Categories", "Characteristics", "Attributes", "Contributions"
]
form.table_rows = []
total_contrib = 0
for cat_name in form.categories:
sub_head = [cat_name, "", "", ""]
form.table_rows += [sub_head]
sum_contribution = 0
for feature_name in form.categories[cat_name]:
feature = Feature.query.filter(
Feature.name == feature_name).first()
case_val = all_case_values_query_for_patient.filter(
CaseValue.feature_id == feature.id).first()
if feature_name == "Age" or feature_name.startswith("Number"):
age = int(case_val.value)
age = age_range_groups(range_groups_ages, age)
form.table_rows += [[
"",
feature_name,
age,
round(case_val.explainer * 100, 3),
]]
else:
form.table_rows += [[
"",
feature_name,
round(case_val.value, 3),
round(case_val.explainer * 100, 3),
]]
sum_contribution += case_val.explainer * 100
sub_head[3] = round(sum_contribution, 3)
total_contrib += sum_contribution
form.table_rows += [["Total", "", "", round(total_contrib, 3)]]
total_score_name_client = round(total_contrib, 3)
prediction_score_list = []
max_patient_id = max([v.case_id for v in all_case_values_query.all()])
for patient_id in range(max_patient_id + 1):
all_case_values_query_for_patient = all_case_values_query.filter(
CaseValue.case_id == patient_id)
if not all_case_values_query_for_patient.all():
continue
age_feature = Feature.query.filter(Feature.name == "Age").first()
age_case_val = all_case_values_query_for_patient.filter(
CaseValue.feature_id == age_feature.id).first()
prediction_score = int(round(age_case_val.prediction, 2) * 100)
prediction_score_list += [prediction_score]
sum_prediction_score = sum(prediction_score_list)
average_default_score = round(sum_prediction_score / (max_patient_id + 1),
2)
return render_template("financial_explan_2_per_client.html",
form=form,
total_score_name_client=total_score_name_client,
average_default_score=average_default_score)
def explanations_per_patient(case_id):
form = FlaskForm(request.form)
form.selected_features = session.get("selected_features", [])
form.categories = json.loads(session.get("categories", "{}"))
form.subdomain = Subdomain.query.get(session.get("subdomain", None))
ranges_for_feature = session.get("ranges_for_feature", {})
range_groups_ages = ranges_for_feature.get("Age", [])
all_case_values_query = CaseValue.query.filter(
CaseValue.user_data_id == session["user_data_id"])
all_case_values_query_for_patient = all_case_values_query.filter(
CaseValue.case_id == case_id)
form.presentation_type = "Proportional to the Total Confidence Score"
if form.validate_on_submit():
form.presentation_type = request.form["presentation_type"]
elif form.is_submitted():
flash("Invalid data", "warning")
form.table_heads = []
sum_explainer = {}
sum_explainer_abs = {}
for cat_name in form.categories:
sum_explainer[cat_name] = 0
sum_explainer_abs[cat_name] = 0
for feature_name in form.categories[cat_name]:
feature = Feature.query.filter(
Feature.name == feature_name).first()
case_val = all_case_values_query_for_patient.filter(
CaseValue.feature_id == feature.id).first()
sum_explainer[cat_name] += case_val.explainer
sum_explainer_abs[cat_name] += abs(case_val.explainer)
form.table_heads += [[
cat_name,
], "Feature Contribution"]
form.table_rows = []
num_of_rows = max([len(form.categories[k]) for k in form.categories])
for cat_name in form.categories:
row_index = 0
for feature_name in form.categories[cat_name]:
if len(form.table_rows) <= row_index:
form.table_rows += [[]]
feature = Feature.query.filter(
Feature.name == feature_name).first()
case_val = all_case_values_query_for_patient.filter(
CaseValue.feature_id == feature.id).first()
if form.presentation_type == "Proportional to the Total Confidence Score":
one_square_val = sum(sum_explainer_abs.values()) / 12
square_num = int(
round(abs(case_val.explainer) / one_square_val, 0))
form.table_rows[row_index] += [[
feature_name, gen_squares_code(square_num)
]]
else:
each_category_squares = 12 / len(form.categories)
value_of_square = int(round(100 / each_category_squares, 0))
sum_explainer_category = sum_explainer_abs[cat_name]
feature_explainer = int(
(case_val.explainer / sum_explainer_category) * 100)
feature_squares = int(
round((feature_explainer / value_of_square), 0))
form.table_rows[row_index] += [[
feature_name,
gen_squares_code(feature_squares)
]]
row_index += 1
for i in range(row_index, num_of_rows):
if len(form.table_rows) <= i:
form.table_rows += [[]]
form.table_rows[i] += [["", ""]]
explainers = sum((list(map(abs, sum_explainer.values()))))
values = list(
map(lambda val: int(round(val * 100 / explainers, 0)),
sum_explainer.values()))
form.value_percent = max(values)
for i, percent in enumerate(values):
form.table_heads[2 * i] += [percent]
age_feature = Feature.query.filter(Feature.name == "Age").first()
age_case_val = all_case_values_query_for_patient.filter(
CaseValue.feature_id == age_feature.id).first()
form.age = int(age_case_val.value)
form.age = age_range_groups(range_groups_ages, form.age)
form.case_id = case_id
form.prediction_score = prediction_score(age_case_val.prediction)
form.predicted = predicted(form.prediction_score)
def check_is_list(val):
return type(val) is list
return render_template(
"explanations_per_patient.html",
form=form,
check_is_list=check_is_list,
enumerate=enumerate,
)
