def main(args):
"""
Runs univariate regression for each of the available metrics.
"""
warnings.filterwarnings("ignore",
category=RuntimeWarning,
module="sklearn")
if args.split_paradigm_score:
folder = f'{args.folder}/split-regression/univariate/'
else:
folder = f'{args.folder}/regression/univariate/'
estimator = LogisticRegression(class_weight='balanced', random_state=42)
cv = StratifiedKFold(n_splits=10, shuffle=True, random_state=42)
for category in categories:
for path, name in projects.items():
df = get_metric_results(args.folder, path, category)
if df is not None:
if args.split_paradigm_score:
for paradigm, scores in split_paradigm_score(
df, args.folder, path, category):
univariate(scores, folder, category, name + paradigm,
estimator, cv, args)
else:
univariate(df, folder, category, name, estimator, cv, args)
if args.split_paradigm_score:
summarise_split_directory(args, 'univariate')
else:
summarise_directory(args, 'univariate')
def main(args):
"""
Calculates statistics of the available fault data.
"""
if args.split_paradigm_score:
folder = f'{args.folder}/split-regression/fault-statistics/'
else:
folder = f'{args.folder}/regression/fault-statistics/'
for category in categories:
statistics = pd.DataFrame(columns=[
'name', 'rows', 'faulty_rows', 'non_faulty_rows',
'percentage_faulty'
])
for path, name in projects.items():
df = get_metric_results(args.folder, path, category)
if df is not None:
if args.split_paradigm_score:
for paradigm, scores in split_paradigm_score(
df, args.folder, path, category):
statistics = fault_statistics(scores, statistics,
category,
name + paradigm)
else:
statistics = fault_statistics(df, statistics, category,
name)
if not statistics.empty:
save_dataframe(statistics, folder, category, False)
def main(args):
"""
Runs multivariate regression on all metrics together.
"""
warnings.filterwarnings("ignore",
category=ConvergenceWarning,
module="sklearn")
if args.split_paradigm_score:
folder = f'{args.folder}/split-regression/multivariate/'
else:
folder = f'{args.folder}/regression/multivariate/'
estimator = LogisticRegression(class_weight='balanced', random_state=42)
cv = StratifiedKFold(n_splits=10, shuffle=True, random_state=42)
for category in categories:
regression_results = pd.DataFrame(columns=[
'name', 'tn', 'fp', 'fn', 'tp', 'r2', 'precision', 'recall', 'mcc'
])
for path, name in projects.items():
df = get_metric_results(args.folder, path, category)
if df is not None:
if args.split_paradigm_score:
for paradigm, scores in split_paradigm_score(
df, args.folder, path, category):
regression_results = multivatiate(
scores, regression_results, category,
name + paradigm, estimator, cv, args)
else:
regression_results = multivatiate(df, regression_results,
category, name,
estimator, cv, args)
if not regression_results.empty:
save_dataframe(regression_results, folder, category, False)
def main(args):
"""
Calculates statistics of the fault data per metric.
"""
if args.split_paradigm_score:
folder = f'{args.folder}/split-regression/fault-metric-statistics/'
else:
folder = f'{args.folder}/regression/fault-metric-statistics/'
for category in categories:
for path, name in projects.items():
df = get_metric_results(args.folder, path, category)
if df is not None:
if args.split_paradigm_score:
for paradigm, scores in split_paradigm_score(
df, args.folder, path, category):
fault_metric_statistics(folder, scores, category,
name + paradigm, args)
else:
fault_metric_statistics(folder, df, category, name, args)
if args.split_paradigm_score:
summarise_split_directory(
args, 'fault-metric-statistics',
['name', 'percentage_faulty', 'percentage_total_faults'])
else:
summarise_directory(
args, 'fault-metric-statistics',
['name', 'percentage_faulty', 'percentage_total_faults'])
def main(args):
"""
Runs univariate and multivariate baseline regression on a control metric.
The control metric is the expected values with a small probability of being incorrect.
"""
warnings.filterwarnings("ignore",
category=ConvergenceWarning,
module="sklearn")
folder = f'{args.folder}/regression/multivariate-baseline-control/'
estimator = LogisticRegression(class_weight='balanced', random_state=42)
cv = StratifiedKFold(n_splits=10, shuffle=True, random_state=42)
for category in categories:
for path, name in projects.items():
df = get_metric_results('baseline', path, category)
if df is not None:
multivariate_baseline_control(df, folder, category, name,
estimator, cv, args)
summarise_directory(args, 'multivariate-baseline-control')
def main(args):
"""
For each of the avialable metrics, runs multivariate regression
on the baseline metric set with one of the metrics added.
"""
warnings.filterwarnings("ignore", category=ConvergenceWarning, module="sklearn")
folder = f'{args.folder}/regression/multivariate-baseline/'
estimator = LogisticRegression(class_weight='balanced', random_state=42)
cv = StratifiedKFold(n_splits=10, shuffle=True, random_state=42)
for category in categories:
for path, name in projects.items():
baseline_df = get_metric_results('baseline', path, category)
df = get_metric_results(args.folder, path, category)
if df is None and 'object' in category:
base_category = 'objectResultsBriand' if 'Briand' in category else 'objectResultsLandkroon'
df = get_metric_results(args.folder, path, base_category)
if baseline_df is not None and df is not None:
multivariate_baseline(baseline_df, df, folder, category, name, estimator, cv, args)
summarise_directory(args, 'multivariate-baseline')
def main(args):
"""
Calculates descriptive statistics of the metric values.
"""
if args.split_paradigm_score:
folder = f'{args.folder}/split-regression/descriptive/'
else:
folder = f'{args.folder}/regression/descriptive/'
for category in categories:
for path, name in projects.items():
df = get_metric_results(args.folder, path, category)
if df is not None:
if args.split_paradigm_score:
for paradigm, scores in split_paradigm_score(
df, args.folder, path, category):
descriptive(scores, folder, category, name + paradigm,
args)
else:
descriptive(df, folder, category, name, args)
if args.split_paradigm_score:
summarise_split_directory(args, 'descriptive', ['name', 'mean', 'std'])
else:
summarise_directory(args, 'descriptive', ['name', 'mean', 'std'])