tslearn.utils.from_tsfresh_dataset

tslearn.utils.from_tsfresh_dataset(X)[source]

Transform a tsfresh-compatible dataset into a tslearn dataset.

Parameters:
X: pandas data-frame

tsfresh-formatted dataset (“flat” data frame, as described there)

Returns:
array, shape=(n_ts, sz, d)

tslearn-formatted dataset. Column order is kept the same as in the original data frame.

Notes

Conversion from/to tsfresh format requires pandas to be installed.

Examples

>>> import pandas as pd
>>> tsfresh_df = pd.DataFrame(columns=["id", "time", "a", "b"])
>>> tsfresh_df["id"] = [0, 0, 0]
>>> tsfresh_df["time"] = [0, 1, 2]
>>> tsfresh_df["a"] = [-1, 4, 7]
>>> tsfresh_df["b"] = [8, -3, 2]
>>> tslearn_arr = from_tsfresh_dataset(tsfresh_df)
>>> tslearn_arr.shape
(1, 3, 2)
>>> tsfresh_df = pd.DataFrame(columns=["id", "time", "a"])
>>> tsfresh_df["id"] = [0, 0, 0, 1, 1]
>>> tsfresh_df["time"] = [0, 1, 2, 0, 1]
>>> tsfresh_df["a"] = [-1, 4, 7, 9, 1]
>>> tslearn_arr = from_tsfresh_dataset(tsfresh_df)
>>> tslearn_arr.shape
(2, 3, 1)
>>> tsfresh_df = numpy.random.randn(10, 1, 16)
>>> from_tsfresh_dataset(
...     tsfresh_df
... )  
Traceback (most recent call last):
...
ValueError: X is not a valid input tsfresh array.