def make_numbers(**options):
from csevo.Table import Table
table_maker = Table()
model = options.get("model")
use_latest = Utils.get_option_as_boolean(options, "use_latest")
debug = Utils.get_option_as_boolean(options, "debug")
table_maker.make_numbers_model_results(model, use_latest, debug)
def process_data_shared(**options):
from csevo.processor.DataProcessor import DataProcessor
output_dir = options.get("output_dir", Macros.data_dir / "models-data")
task = options.get("task")
years = Utils.get_option_as_list(options, "years")
eval_settings = Utils.get_option_as_list(options, "eval_settings")
dp = DataProcessor()
dp.process_shared(output_dir, years, eval_settings, task)
return
def prepare_model_local(**options):
from csevo.ml.LocalRunner import LocalRunner
work_dir = Path(options.get("work_dir", Macros.data_dir / "models-work"))
model = options["model"]
use_latest = Utils.get_option_as_boolean(options, "use_latest")
debug = Utils.get_option_as_boolean(options, "debug")
cross_proj = Utils.get_option_as_boolean(options, "cross_proj")
runner = LocalRunner(work_dir)
runner.prepare_model(model, use_latest, debug, cross_proj)
return
def store_repo_results(**options):
from csevo.collector.Database import Database
from csevo.collector.DataCollector import DataCollector
local = Utils.get_option_as_boolean(options, "local")
force_update = Utils.get_option_as_boolean(options, "force_update")
repos_results_dir = Path(
options.get("repos_results_dir", Macros.repos_results_dir))
db = Database(local=local)
dc = DataCollector(database=db)
dc.store_repo_results(repos_results_dir, force_update=force_update)
return
def make_plots(**options):
from csevo.Plot import Plot
which = Utils.get_option_as_list(options, "which")
plot_maker = Plot()
plot_maker.make_plots(which, options)
return
def make_tables(**options):
from csevo.Table import Table
which = Utils.get_option_as_list(options, "which")
table_maker = Table()
table_maker.make_tables(which, options)
return
def collect_model_results(**options):
from csevo.collector.ModelResultsCollector import ModelResultsCollector
collector = ModelResultsCollector()
model = options.get("model")
task = options.get("task")
re_eval = Utils.get_option_as_boolean(options, "re_eval")
collector.collect_results(model, task, re_eval)
def split_project_data(**options):
from csevo.filter.DataSpliter import DataSpliter
spliter = DataSpliter()
task = options.get("task")
debug = Utils.get_option_as_boolean(options, "debug")
spliter.project_data_split(task, debug)
return
def make_plots(self, which, options: dict):
for item in which:
if item == "draft-learning-curve":
# TODO: outdated (->remove)
training_log_path = Path(options.get("training-log-path"))
output_name = options.get("output-name")
self.make_plot_draft_learning_curve(training_log_path, output_name)
elif item == "models-results-metrics-dist":
task = options["task"]
models = Utils.get_option_as_list(options, "models", self.TASK_2_MODELS.get(task))
metrics = Utils.get_option_as_list(options, "metrics", self.TASK_2_METRICS.get(task))
exps = Utils.get_option_as_list(options, "exps", self.EXPS)
self.plot_models_results_metrics_dist(task, models, metrics, exps)
elif item == "models-results-variance-dist":
task = options["task"]
models = Utils.get_option_as_list(options, "models", self.TASK_2_MODELS.get(task))
metrics = Utils.get_option_as_list(options, "metrics", self.TASK_2_METRICS.get(task))
exps = Utils.get_option_as_list(options, "exps", self.EXPS)
self.plot_models_results_variance_dist(task, models, metrics, exps)
elif item == "num-data-evolution":
self.plot_num_data_evolution(
Utils.get_option_as_list(options, "years", self.EVO_YEARS),
)
else:
self.logger.warning(f"Unknown plot {item}")
# end if
# end for
return
def run_models(**options):
from csevo.ml.TACCRunner import TACCRunner
work_dir = Path(options.get("work_dir", Macros.data_dir / "models-work"))
mode = options.get("mode", Macros.train)
models = Utils.get_option_as_list(options, "models")
exps = Utils.get_option_as_list(options, "exps")
trials = Utils.get_option_as_list(options, "trials")
timeout = options.get("timeout")
beg = options.get("beg", 0)
cnt = options.get("cnt", -1)
local = Utils.get_option_as_boolean(options, "local")
runner = TACCRunner(work_dir)
if not local:
runner.run_models(mode, models, exps, trials, timeout, beg, cnt)
else:
runner.run_models_local(mode, models, exps, trials, timeout, beg, cnt)
return
def split_dataset(**options):
from csevo.collector.Database import Database
from csevo.processor.DatasetSplitter import DatasetSplitter
local = Utils.get_option_as_boolean(options, "local")
db = Database(local=local)
ds = DatasetSplitter(database=db)
ds.split_dataset()
return
def split_projects(**options):
from csevo.filter.DataSpliter import DataSpliter
spliter = DataSpliter()
random_seed = options.get("random_seed")
debug = Utils.get_option_as_boolean(options, "debug")
if debug:
method_file = Macros.data_dir / "latest-debug" / "method-data.json"
else:
method_file = Macros.data_dir / "latest" / "method-data.json"
spliter.split_project(method_file, random_seed, debug)
def prepare_model(**options):
from csevo.ml.TACCRunner import TACCRunner
work_dir = Path(options.get("work_dir", Macros.data_dir / "models-work"))
model = options["model"]
year = options["year"]
eval_setting = options["eval_setting"]
debug = Utils.get_option_as_boolean(options, "debug")
runner = TACCRunner(work_dir)
runner.prepare_model(model, year, eval_setting, debug)
return
def collect_data(**options):
from csevo.collector.DataCollector import DataCollector
project_urls_file = Path(
options.get("project_urls_file",
Macros.data_dir / "projects-github-dpcom.txt"))
skip_collected = Utils.get_option_as_boolean(options, "skip_collected")
beg = options.get("beg")
cnt = options.get("cnt")
dc = DataCollector()
dc.collect_projects(project_urls_file,
skip_collected=skip_collected,
beg=beg,
cnt=cnt)
return