ramjet.photometric_database.lightcurve_database

Code for a base generalized database for photometric data to be subclassed.

Module Contents

class LightcurveDatabase(data_directory='data')

Bases: abc.ABC

A base generalized database for photometric data to be subclassed.

__init__(self, data_directory='data')

log_dataset_file_names(self, dataset: tf.data.Dataset, dataset_name: str)

Saves the names of the files used in a dataset to a CSV file in the trail directory.

static normalize_log_0_to_1(lightcurve: np.ndarray)

Normalizes from 0 to 1 on the logarithm of the lightcurve.

static normalize(lightcurve: np.ndarray)

Normalizes light curve using quantile. quantile_10 (10%) at -1 and quantile_90 (90%) at 1.

static shuffle_in_unison(a, b, seed=None)

Shuffle two arrays in unison.

static remove_random_values(lightcurve: np.ndarray)

Removes random values from the lightcurve.

get_ratio_enforced_dataset(self, positive_training_dataset: tf.data.Dataset, negative_training_dataset: tf.data.Dataset, positive_to_negative_data_ratio: float)

Generates a dataset with an enforced data ratio.

static repeat_dataset_to_size(dataset: tf.data.Dataset, size: int)

Repeats a dataset to make it a desired length.

is_positive(self, example_path)

Checks if an example contains a microlensing event or not.

Parameters:example_path – The path to the example to check. Returns:Whether or not the example contains a microlensing event.

static make_uniform_length(example: np.ndarray, length: int, randomize: bool = True, seed: int = None)

Makes the example a specific length, by clipping those too large and repeating those too small.

get_training_and_validation_datasets_for_file_paths(self, example_paths: Union[Iterable[Path], Callable[[], Iterable[Path]]])

Creates training and validation datasets from a list of all file paths. The database validation ratio is used to determine the size of the split.

Parameters:example_paths – The total list of file paths. Returns:The training and validation datasets.

static extract_shuffled_chunk_and_remainder(array_to_extract_from: Union[List, np.ndarray], chunk_ratio: float, chunk_to_extract_index: int = 0)

Shuffles an array, extracts a chunk of the data, and returns the chunk and remainder of the array.

Parameters:

• array_to_extract_from – The array to process.
• chunk_ratio – The number of equal size chunks to split the array into before extracting one.
• chunk_to_extract_index – The index of the chunk to extract out of all chunks.

Returns:

The chunk which is extracted, and the remainder of the array excluding the chunk.

static padded_window_dataset_for_zipped_example_and_label_dataset(dataset: tf.data.Dataset, batch_size: int, window_shift: int, padded_shapes: Tuple[List, List])

Takes a zipped example and label dataset, and converts it to padded batches, where each batch uses overlapping examples based on a sliding window.

Parameters:

• dataset – The zipped example and label dataset.
• batch_size – The size of the batches to produce.
• window_shift – The shift of the moving window between batches.
• padded_shapes – The output padded shape.

Returns:

The padded window dataset.

static window_dataset_for_zipped_example_and_label_dataset(dataset: tf.data.Dataset, batch_size: int, window_shift: int)

Takes a zipped example and label dataset, and converts it to batches, where each batch uses overlapping examples based on a sliding window.

Parameters:

• dataset – The zipped example and label dataset.
• batch_size – The size of the batches to produce.
• window_shift – The shift of the moving window between batches.

Returns:

The window dataset.

clear_data_directory(self)

Empties the data directory.

create_data_directories(self)

Creates the data directories to be used by the database.

static paths_dataset_from_list_or_generator_factory(list_or_generator_factory: Union[Iterable[Path], Callable[[], Iterable[Path]]])

Produces a dataset from either the examples path list or example paths factory to strings.

Parameters:list_or_generator_factory – The list or generator factory. Returns:The new path generator.