K-Medoids Clustering

Related Cubes

Calculation Parameters

  • CPUs (cpu_count) type: integer: The number of CPUs to run this cube with
    Default: 1 , Min: 1, Max: 128
  • Cube Metrics (cube_metrics) type: string: Set of metrics to be collected

    Choices: cpu, disk, memory, network
  • Use Diagnostics (debug_mode) type: boolean:
    Default: False
  • Temporary Disk Space (MiB) (disk_space) type: decimal: The minimum amount of disk space in MiB (1048576 B) this cube requires. Due to overhead, request a couple hundred MiB more than required.
    Default: 5120.0 , Min: 128.0, Max: 8589934592
  • GPUs (gpu_count) type: integer: The number of GPUs to run this cube with
    Default: 0 , Max: 16
  • Medoid Initialization Method (init_kmedoids) type: string: From scikit-learn-extra docs: Specify medoid initialization method. ‘random’ selects n_clusters elements from the dataset. ‘heuristic’ picks the n_clusters points with the smallest sum distance to every other point. ‘k-medoids++’ follows an approach based on k-means++_, and in general, gives initial medoids which are more separated than those generated by the other methods. ‘build’ is a greedy initialization of the medoids used in the original PAM algorithm. Often ‘build’ is more efficient but slower than other initializations on big datasets and it is also very non-robust, if there are outliers in the dataset, use another initialization.
    Default: heuristic
    Choices: random, heuristic, k-medoids++, build
  • Instance Tags (instance_tags) type: string: Only run on machines with matching tags (comma separated)
    Default: “”
  • Instance Type (instance_type) type: string: The type of instance that this cube needs to be run on
  • None (matrix_input_file) type: file_in:
  • Maximum K-Medoids Iterations (max_iter) type: integer:
    Default: 100000
  • Memory (MiB) (memory_mb) type: decimal: The minimum amount of memory in MiBs (1048576 B) this cube requires. Due to overhead, request a couple hundred MiB more than required.
    Default: 1800 , Min: 256.0, Max: 8589934592
  • Algorithm (method) type: string: Alternate is faster, pam is more accurate
    Default: pam
    Choices: alternate, pam
  • Distance Metric (metric) type: string: What distance metric to use.
    Default: precomputed
  • Metric Period (metric_period) type: decimal: How often to sample metrics, in seconds
    Default: 60
    Choices: 1, 5, 10, 30, 60, 120, 180, 240, 300, Min: 1, Max: 300
  • Number of K-Medoids Clusters (n_clusters) type: integer:
  • Output Similarity Matrix (output_similarity_matrix) type: boolean:
    Default: False
  • Thread limit per CPU (pids_per_cpu_limit) type: integer: The number of threads per CPU
    Default: 32
  • Random State (random_state) type: integer: Specify random state for the random number generator. Used to initialise medoids when init=’random’.
  • None (row_label_input_file) type: file_in:
  • Shared Memory (MiB) (shared_memory_mb) type: decimal: The amount of shared memory to allow a container to address
    Default: 64
  • Similarity Matrix Filename (similarity_matrix_filename) type: string:
    Default: clustering_similarity_matrix.txt
  • Spot policy (spot_policy) type: string: Control cube placement on spot market instances
    Default: Prohibited
    Choices: Allowed, Preferred, NotPreferred, Prohibited, Required
  • None (use_matrix_input_file) type: boolean:
    Default: False

Field parameters

  • Cluster ID Field (cluster_id_field) type: Field Type: String: The name for the field that will contain the unique cluster ID.
    Default: Cluster ID
  • Cluster Method Field (cluster_method_field) type: Field Type: String: Field name for passing the clustering method to the floe report.
    Default: Cluster Method
  • Cluster Parameters Field (cluster_parameters_field) type: Field Type: String: Field name for passing the cluster parameters to the Floe report.
    Default: Parameters
  • None (coord_list_field) type: Field Type: IntVec:
    Default: coord_list_field
  • Extended Log Field (ext_log_field) type: Field Type: StringVec: Message extended log field
    Default: Extended Log Field
  • None (is_core) type: Field Type: Bool:
    Default: is_core
  • Log Field (log_field) type: Field Type: String: The field to store messages to floe report
    Default: Log Field
  • None (matrix_size_field) type: Field Type: Int:
    Default: matrix_size
  • None (mol_field) type: Field Type: Chem.Mol:
  • Similarity Score Field (score_field) type: Field Type: Float: Name for the field that stores fingerprint similarity scores.
    Default: similarity_score
  • None (score_list_field) type: Field Type: FloatVec:
    Default: score_list_field
  • None (x_field) type: Field Type: Int:
    Default: x
  • None (y_field) type: Field Type: Int:
    Default: y

Hardware Parameters

Machine hardware requirements

  • Memory (MiB) (memory_mb) type: decimal: The minimum amount of memory in MiBs (1048576 B) this cube requires. Due to overhead, request a couple hundred MiB more than required.
    Default: 1800 , Min: 256.0, Max: 8589934592
  • Shared Memory (MiB) (shared_memory_mb) type: decimal: The amount of shared memory to allow a container to address
    Default: 64
  • Thread limit per CPU (pids_per_cpu_limit) type: integer: The number of threads per CPU
    Default: 32
  • Temporary Disk Space (MiB) (disk_space) type: decimal: The minimum amount of disk space in MiB (1048576 B) this cube requires. Due to overhead, request a couple hundred MiB more than required.
    Default: 5120.0 , Min: 128.0, Max: 8589934592
  • GPUs (gpu_count) type: integer: The number of GPUs to run this cube with
    Default: 0 , Max: 16
  • CPUs (cpu_count) type: integer: The number of CPUs to run this cube with
    Default: 1 , Min: 1, Max: 128
  • Instance Type (instance_type) type: string: The type of instance that this cube needs to be run on
  • Spot policy (spot_policy) type: string: Control cube placement on spot market instances
    Default: Prohibited
    Choices: Allowed, Preferred, NotPreferred, Prohibited, Required
  • Instance Tags (instance_tags) type: string: Only run on machines with matching tags (comma separated)
    Default: “”

Metrics Parameters

Cube Metric Parameters

  • Metric Period (None) type: decimal: How often to sample metrics, in seconds
    Default: 60
    Choices: 1, 5, 10, 30, 60, 120, 180, 240, 300, Min: 1, Max: 300
  • Cube Metrics (None) type: string: Set of metrics to be collected

    Choices: cpu, disk, memory, network

Parallel K-Medoids Clustering

The parallel version adds these extra parameters.

  • Number of messages to distribute at a time (item_count) type: integer: The maximum number of messages to bundle together for a parallel cube.
    Default: 1 , Min: 1, Max: 65535
  • Maximum Failures (max_failures) type: integer: The maximum number of times to attempt processing a work item
    Default: 10 , Min: 1, Max: 100
  • Autoscale this Cube (autoscale) type: boolean: If True, let Orion manage the parallelism of this Cube
    Default: True
  • Maximum number of Cubes (max_parallel) type: integer: The maximum number of concurrently running copies of this Cube
    Default: 1000 , Min: 1
  • Minimum number of Cubes (min_parallel) type: integer: The minimum number of concurrently running copies of this Cube
    Default: 0