deepof.post_hoc
Data structures and functions for analyzing supervised and unsupervised model results.
Functions
Align kinematics with unsupervised labels. |
|
|
Annotate time chunks produced after change-point detection using the unsupervised pipeline. |
|
Split a dataset into training and testing sets, grouped by video. |
|
Extract summary statistics from a chunked dataset using seglearn. |
|
Compute UMAP embeddings for visualization purposes. |
|
Compute the steady state of each transition matrix provided in a dictionary. |
Compute the transition matrices specific to each condition. |
|
|
Compute the distance between the embeddings of two conditions, using the specified aggregation method. |
|
Compute the population of each cluster across conditions. |
|
Compute SHAP feature importance for models mapping chunk_stats to cluster assignments. |
Fit a global model to the normal embeddings. |
|
|
Aggregate the embeddings of a set of videos, using the specified aggregation method. |
|
Compute how much each animal spends on each cluster. |
|
Compute the transitions between states in a state sequence. |
|
Recluster the data using a HMM-based approach. |
|
Select a time bin and filters all relevant objects (embeddings, soft_counts, breaks, and supervised annotations). |
|
Compute the distance between the embeddings of two conditions, using the specified aggregation method. |
|
Train supervised models to detect clusters from kinematic features. |