Encode text strings to embeddings via OpenAI API. Parameters ---------- X : list of str Text strings to encode.

Fit the Anomaly Transformer on time series data. Parameters ---------- X : array-like of shape (n_timestamps,) or (n_timestam

Fit the frame autoencoder and score the training clips. Parameters ---------- X : list Audio clips as file paths,

Fit detector. y is ignored in unsupervised methods. Parameters ---------- X : numpy array of shape (n_samples, n_features)

Get benchmark results. Parameters ---------- benchmark : str Benchmark name, or 'all' for everything. Re

Return the filesystem directory containing a packaged skill. Parameters ---------- skill_name : str Name of the packaged ski

List algorithms supporting a given data type.

List algorithms with a given status.

Set the parameters of this estimator. The method works on simple estimators as well as on nested objects (such as pipelines). The latt

Suggest an alternative detector different from the current one. Resolution order: 1. First entry in the current plan's ``alternatives`` list

Convert edge_index to scipy sparse CSR adjacency. Parameters ---------- edge_index : np.ndarray of shape (2, n_edges) COO edge li

Compute hindsight metrics from held-out labels.

Utility function to calculate the weighted Pearson correlation of two samples. See https://stats.stackexchange.com/questions/221246/such-thin

FILTER class for Filtering based thresholders. Use the filtering based methods to evaluate a non-parametric means to threshold scores g

META class for Meta-modelling thresholder. Use a trained meta-model to evaluate a non-parametric means to threshold scores generated by

Internal function to calculate for dissimilarity in a sequence of sets.

Computes the binding probabilities of a given affinity matrix. Parameters ---------- A : array, shape (n_samp

Return True if `cls.__init__` declares an explicit `random_state` parameter. A class is considered to accept ``random_state`` only when the param

Mean per-frame score within each clip.

The average path length in a n_samples iTree, which is equal to the average path length of an unsuccessful BST search since the latter ha

Computes the binding probabilities of a given affinity matrix. Parameters ---------- A : array, shape (n_samp

Walk ADEngine + InvestigationState via multi-modality dry runs. Returns a tuple of: (engine_attrs, state_attrs, valid_dict_keys, engine_s

Create (input, target) sliding-window pairs. Parameters ---------- X : np.ndarray of shape (n_timestamps, n_channels)

Computes the outlier scores. Parameters ---------- X : array-like, shape (n_samples, n_features) The inpu

The internal function to calculate the outlier scores based on the bins and histograms constructed with the training data. The program is opti

The internal function to calculate the outlier scores based on the bins and histograms constructed with the training data. The program is opti

Compatibility wrapper for sklearn.utils.check_array. Handles force_all_finite -> ensure_all_finite rename in sklearn 1.6+

Internal function to assert univariate data

Check that openai is installed and exposes the OpenAI client.

Combine per-modality scores.

Compute the k-Shape centroid for a set of z-normalized, shift-aligned cluster members. The centroid maximizes total similarity. It is the ei

Compute per-timestamp anomaly scores. Parameters ---------- X : np.ndarray of shape (n_timestamps, n_channels) Va

Create anomalous view by injecting synthetic anomalies. Following the CONAD paper: perturb a subset of node attributes by swapping with ra

Create overlapping windows preserving the channel dimension. Parameters ---------- X : np.ndarray, shape (n_timestamps, n_channels) w

Performs a binary search to get affinities in such a way that each conditional Gaussian has the same perplexity. Then returns the affi

Parse a `run_detection` JSON result back into a dict. Converts the list-form ``scores_train`` / ``labels_train`` / ``scores_test`` / ``labels

Return (frames, clip_idx) over all clips in X.

Internal function to generate data samples. Parameters ---------- n_inliers : int The number of inliers. n_outliers :

Computes the alpha neighbourhood points. Parameters ---------- dist_matrix : array-like, shape (n_samples, n_features)

Helper function for getting outlier scores on test data X (note: model must already be fit) Parameters ---------- X

Internal function to decide how to process passed in distance array Parameters ---------- dist_arr : numpy array of shape (n_

Route a loaded top-level object through envelope/legacy handlers. Shared by the non-fall-through path and by the post-`compat_load` path; the

Map a Python subpackage name to its Claude Code install dirname.

Map per-window per-timestamp scores to full time series scores. For each window, per-position scores are assigned to the corresponding global

Collapse a waveform array to 1-D mono. 1-D arrays pass through. For 2-D arrays the channel axis is the smaller of the two dimensions (audio h

Private function used to partition estimators between jobs. See sklearn/ensemble/base.py for more information.

Run the trained LSTM model on input windows. Parameters ---------- model : _LSTMModel inputs : np.ndarray of shape (n

Fit preprocessing and transform embeddings.

Calculate the mean Cook's distances for each feature. Parameters ---------- X : numpy array of shape (n_samples, n_features) The

Render a list of (name, algo) tuples as a markdown bullet list. Each bullet contains: name, full_name, complexity, best_for, avoid_when, requ

Resolve the device string to a torch.device.

Shared install path for both `pyod install skill` and `pyod-install-skill`. Parameters mirror the argparse surface of both CLIs. Returns a shel

Score each subsequence by SBD to nearest centroid. Parameters ---------- subs : np.ndarray of shape (n_subs, length)

Compute anomaly scores for a batch of windows. Score(t) = softmax(-AssDis(t)) * ||x_t - x_hat_t||^2 Parameters ---------- model : _A

Return deduplicated (name, algo) tuples whose data_types include any modality. Used by both single-modality renderers and the combined text/image

Compute spectral residual saliency for a 1-D signal. Parameters ---------- x : np.ndarray of shape (n_timestamps,)

Return a 1-D mono waveform for one input item. Accepts a file path (str), a waveform array, or a ``(waveform, sample_rate)`` tuple. Multi-cha

Update the Matrix Profile for column j. Converts QT values to z-normalized distances and updates the profile wherever the new distance is sma

Wrap-around Quasi-Monte Carlo discrepancy method

Computes the dissimilarity matrix of a given dataset. Parameters ---------- X : array-like, shape (n_samples, n_featu

Z-normalize each row of a window matrix.

Increase contamination by `_CONTAMINATION_INCREASE_FACTOR`, capped at `_CONTAMINATION_MAX`.

Handle structured feedback dict. Mutates `state` in place to apply the feedback action, then resets detection-side fields (``results``, ``con

Calls the OpenAI GPT-4o API with the provided prompt and returns the response. Parameters ---------- prompt

Compute three diagnostic quality metrics for a detection run. Each metric diagnoses one independent failure mode and drives one branch of `AD

Predict raw anomaly score of X using the fitted detector. The anomaly score of an input sample is computed based on different detecto

Predict raw anomaly score of X using the fitted detector. The anomaly score of an input sample is computed based on different detecto

Predict raw anomaly score of X using the fitted detector. The anomaly score of an input sample is computed based on different detecto

Predict raw anomaly scores for a test time series. Parameters ---------- X : array-like of shape (n_timestamps,) or (n_timest

Predict raw anomaly score of X using the fitted detector. The anomaly score of an input sample is computed based on different detecto

Predict raw anomaly score of X using the fitted detector. For consistency, outliers are assigned with larger anomaly scores. Parameter

Convert raw input to numeric embeddings. Parameters ---------- X : list or array-like Raw input data. ba

Encode audio clips to a (n_samples, 74) feature matrix. Parameters ---------- X : list Audio clips as file paths,

Find the euclidean distance between two vectors or between a vector and a collection of vectors. Parameters ---------- v1 : list

Evaluate the deep learning model. Parameters ---------- data_loader : torch.utils.data.DataLoader The da

Yield every *.md file under pyod/skills/ recursively.

Fit detector. y is ignored in unsupervised methods. Parameters ---------- X : numpy array of shape (n_samples, n_features)

Fit detector. y is ignored in unsupervised methods. Parameters ---------- X : numpy array of shape (n_samples, n_features)