Module jidenn.data
This module contains all the necessary toolkit to:
- convert a ROOT file into a Tensorflow
tf.data.Dataset
object - flatten the dataset from events to individal jets
- pick the subsample of desired variables and apply cuts on them
- remap the labels to a classes used as targets in the training
- resample the dataset to balance the classes
- combine datasets created from different ROOT files
- construct desired input features and targets for the training
- batch and prefetch the dataset for training
These functionalities can be configured using the jidenn.config
module.
Expand source code
"""
This module contains all the necessary toolkit to:
- convert a ROOT file into a Tensorflow `tf.data.Dataset` object
- flatten the dataset from events to individal jets
- pick the subsample of desired variables and apply cuts on them
- remap the labels to a classes used as targets in the training
- resample the dataset to balance the classes
- combine datasets created from different ROOT files
- construct desired input features and targets for the training
- batch and prefetch the dataset for training
These functionalities can be configured using the `jidenn.config` module.
![data_off](images/data_prep_off.png)
![data_on](images/data_prep_on.png)
"""
Sub-modules
jidenn.data.JIDENNDataset
-
Module containing the
jidenn.data.JIDENNDataset
dataclass that is a wrapper for a TensorFlow dataset that allows for easy adding and processing of dataset … jidenn.data.ROOTDataset
-
Module for reading ROOT files and converting them to Tensorflow
tf.RaggedTensor
ortf.Tensor
objects. The module contains thejidenn.data.ROOTDataset
class … jidenn.data.TrainInput
-
Module containing classes that create the various types of input variables for all the neural networks. Each type of input variables is a subclass of …
jidenn.data.convert_h5
jidenn.data.data_info
jidenn.data.four_vector_transform
-
This module contains functions to transform 4-vectors between different representations. It also contains a functions for various 4-vector operations …
jidenn.data.get_dataset
-
Module containing a single function to load and preprocess a dataset from a list of files into
jidenn.data.JIDENNDataset
objects, perform the preprocessing … jidenn.data.string_conversions
-
This module contains functions to convert strings to objects that can be evaluated on a sample, without using
eval
. The two commonly operations …