tropea_clustering.helpers.reshape_from_dnt¶

tropea_clustering.helpers.reshape_from_dnt(input_data, delta_t)[source]¶

Reshapes the input data from traditional from scikit format.

Takes the array containing the univariate time-series data in the (n_dims, n_particles, n_frames) format and reshapes it in the format required by scikit-learn (-1, delta_t * n_dims).

Parameters:

input_data (np.ndarray of shape (n_dims, n_particles, n_frames)) – The data to cluster in the traditional shape.
delta_t (int) – Length of the signal sequence - the analysis time resolution.

Returns:

reshaped_data – The data to cluster in the scikit-required shape.

Return type:

np.ndarray of shape (n_particles * n_seq, delta_t * n_dim)

Example

import numpy as np
from tropea_clustering import helpers

# Select time resolution
delta_t = 5

# Create random input data
np.random.seed(1234)
n_dims = 2
n_particles = 5
n_frames = 1000

input_data = np.random.rand(n_dims, n_particles, n_frames)

# Create input array with the scikit-required shape
reshaped_input_data = helpers.reshape_from_dnt(
    input_data, delta_t,
)