Task Manager (starclass.TaskManager)

class starclass.TaskManager(todo_file, cleanup=False, readonly=False, overwrite=False, classes=None, load_into_memory=False, backup_interval=10000)

Bases: object

A TaskManager which keeps track of which targets to process.

Code author: Rasmus Handberg <rasmush@phys.au.dk>

__init__(todo_file, cleanup=False, readonly=False, overwrite=False, classes=None, load_into_memory=False, backup_interval=10000)

Initialize the TaskManager which keeps track of which targets to process.

Parameters:

• todo_file (str) – Path to the TODO-file.

• cleanup (bool) – Perform cleanup/optimization of TODO-file before doing initialization. Default=False.

• overwrite (bool) – Overwrite any previously calculated results. Default=False.

• classes (Enum) – Possible stellar classes. This is only used for for translating saved stellar classes in the other_classifiers table into proper enums.

• load_into_memory (bool) – Create a in-memory copy of the entire TODO-file, and work of this copy to speed up queries. Will result in larger memory use. Default=True.

• backup_interval (int) – Save in-memory copy of database to disk after this number of results saved by save_results(). Default=10000.

Raises:

FileNotFoundError – If TODO-file could not be found.

Code author: Rasmus Handberg <rasmush@phys.au.dk>

assign_final_class(tset, data_dir=None)

Assing final classes based on all starclass results.

This will create a new column in the todolist table named “final_class”.

Parameters:

• tset (TrainingSet) – Training-set used.

• data_dir (str, optional) – Data directory to load models from.

Code author: Rasmus Handberg <rasmush@phys.au.dk>

backup()

Save backup of todo-file to disk. This only has an effect when load_into_memory is enabled.

Code author: Rasmus Handberg <rasmush@phys.au.dk>

close()

Close TaskManager and all associated objects.

get_number_tasks(classifier=None)

Get number of tasks to be processed.

Parameters:

classifier (str, optional) – Constrain to tasks missing from this classifier.

Returns:

Number of tasks due to be processed.

Return type:

int

Code author: Rasmus Handberg <rasmush@phys.au.dk>

get_task(priority=None, classifier=None, change_classifier=True, chunk=1, ignore_existing=False)

Get next task to be processed.

Parameters:

• priority (integer)

• classifier (string) – Classifier to get next task for. If no tasks are available for this classifier, and change_classifier=True, a task for another classifier will be returned.

• change_classifier (boolean) – Return task for another classifier if there are no more tasks for the provided classifier. Default=True.

• chunk (int, optional) – Chunk of tasks to return. Default is to not chunk (=1).

Returns:

List of dictionaries of settings for tasks.

If no tasks are found None is returned.

Return type:

list or None

Code author: Rasmus Handberg <rasmush@phys.au.dk>

moat_clear()

Clear Mother Of All Tables (MOAT).

Code author: Rasmus Handberg <rasmush@phys.au.dk>

moat_create(classifier, columns)

moat_query(classifier, priority)

Query Mother Of All Tables (MOAT) for cached features.

Parameters:

• classifier (str)

• priority (int)

Returns:

Dictionary with features stores in MOAT.

Return type:

dict

Code author: Rasmus Handberg <rasmush@phys.au.dk>

save_results(results)

Save results, or list of results, to TODO-file.

Parameters:

results (list or dict) – Dictionary of results and diagnostics.

Raises:

ValueError – If attempting to save results from multiple different training sets.

Code author: Rasmus Handberg <rasmush@phys.au.dk>

save_settings()

Save settings to TODO-file and create method-specific columns in diagnostics_corr table.

Code author: Rasmus Handberg <rasmush@phys.au.dk>

start_task(tasks)

Mark tasks as STARTED in the TODO-list.

Parameters:

tasks (list or dict) – Task or list of tasks coming from get_tasks().

Code author: Rasmus Handberg <rasmush@phys.au.dk>