Fixes typos and references (#657)

charavelg · web-flow · commit 548367480b5f · 2026-03-23T12:56:51.000+01:00
diff --git a/docs/backend.rst b/docs/backend.rst
@@ -3,7 +3,7 @@
 Backend selection and use
 =========================
 
-`tslearn` proposes different backends (`NumPy` and `PyTorch`)
+``tslearn`` proposes different backends (`NumPy` and `PyTorch`)
 to compute time series metrics such as `DTW` and `Soft-DTW`.
 The `PyTorch` backend can be used to compute gradients of
 metric functions thanks to automatic differentiation. The `PyTorch`
diff --git a/docs/conf.py b/docs/conf.py
@@ -96,7 +96,6 @@ def matplotlib_svg_scraper(*args, **kwargs):
 
 # General information about the project.
 project = u'tslearn'
-copyright = u'2025, Romain Tavenard'
 
 # The version info for the project you're documenting, acts as replacement for
 # |version| and |release|, also used in various other places throughout the
diff --git a/docs/examples/classification/plot_early_classification.py b/docs/examples/classification/plot_early_classification.py
@@ -7,11 +7,11 @@
 
 Early classifiers are implemented in the 
 :mod:`tslearn.early_classification` module and in this example 
-we use the method from [1].
+we use the method from [1]_.
 
 
-[1] A. Dachraoui, A. Bondu & A. Cornuejols. Early classification of time
-series as a non myopic sequential decision making problem. ECML/PKDD 2015
+.. [1] A. Dachraoui, A. Bondu & A. Cornuejols. Early classification of time
+  series as a non myopic sequential decision making problem. ECML/PKDD 2015
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/classification/plot_svm.py b/docs/examples/classification/plot_svm.py
@@ -7,13 +7,14 @@
 support vector classification.
 
 This metric is defined in the :ref:`tslearn.metrics <mod-metrics>` module and
-explained in details in [1].
+explained in details in [1]_.
 
 In this example, a `TimeSeriesSVC` model that uses GAK as kernel is fit and the
 support vectors for each class are reported.
 
 
-[1] M. Cuturi, "Fast global alignment kernels," ICML 2011.
+.. [1] M. Cuturi, "Fast global alignment kernels",
+  ICML 2011.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/clustering/plot_barycenter_interpolate.py b/docs/examples/clustering/plot_barycenter_interpolate.py
@@ -5,7 +5,7 @@
 
 This example presents the weighted Soft-DTW time series barycenter method.
 
-Soft-DTW [1] is a differentiable loss function for Dynamic Time Warping,
+Soft-DTW [1]_ is a differentiable loss function for Dynamic Time Warping,
 allowing for the use of gradient-based algorithms. The barycenter corresponds
 to the time series that minimizes the sum of the distances between that time
 series and all the time series from a dataset. It is thus an optimization
@@ -18,8 +18,8 @@
 time series the barycenter is, the higher the weight for this time series
 is.
 
-[1] M. Cuturi and M. Blondel, "Soft-DTW: a Differentiable Loss Function for
-Time-Series". International Conference on Machine Learning, 2017.
+.. [1] M. Cuturi and M. Blondel, "Soft-DTW: a Differentiable Loss Function for
+  Time-Series". International Conference on Machine Learning, 2017.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/clustering/plot_barycenters.py b/docs/examples/clustering/plot_barycenters.py
@@ -18,25 +18,25 @@
 * *DTW Barycenter Averaging (DBA)* is an iteratively refined barycenter,
   starting out with a (potentially) bad candidate and improving it
   until convergence criteria are met. The optimization can be accomplished
-  with (a) expectation-maximization [1] and (b) stochastic subgradient
-  descent [2]. Empirically, the latter "is [often] more stable and finds better
-  solutions in shorter time" [2].
+  with (a) expectation-maximization [1]_ and (b) stochastic subgradient
+  descent [2]_. Empirically, the latter "is [often] more stable and finds better
+  solutions in shorter time" [2]_.
 * *Soft-DTW barycenter* uses a differentiable loss function to iteratively
-  find a barycenter [3]. The method itself and the parameter
+  find a barycenter [3]_. The method itself and the parameter
   :math:`\\gamma=1.0` is described in more detail in the section on
   :ref:`DTW<dtw>`. There is also a dedicated
   :ref:`example<sphx_glr_auto_examples_clustering_plot_barycenter_interpolate.py>`
   available.
 
-[1] F. Petitjean, A. Ketterlin & P. Gancarski. A global averaging method for
-dynamic time warping, with applications to clustering. Pattern Recognition,
-Elsevier, 2011, Vol. 44, Num. 3, pp. 678-693.
+.. [1] F. Petitjean, A. Ketterlin & P. Gancarski. A global averaging method for
+  dynamic time warping, with applications to clustering. Pattern Recognition,
+  Elsevier, 2011, Vol. 44, Num. 3, pp. 678-693.
 
-[2] D. Schultz & B. Jain. Nonsmooth Analysis and Subgradient Methods for
-Averaging in Dynamic Time Warping Spaces. Pattern Recognition, 74, 340-358.
+.. [2] D. Schultz & B. Jain. Nonsmooth Analysis and Subgradient Methods for
+  Averaging in Dynamic Time Warping Spaces. Pattern Recognition, 74, 340-358.
 
-[3] M. Cuturi & M. Blondel. Soft-DTW: a Differentiable Loss Function for
-Time-Series. ICML 2017.
+.. [3] M. Cuturi & M. Blondel. Soft-DTW: a Differentiable Loss Function for
+  Time-Series. ICML 2017.
 """
 
 # Author: Romain Tavenard, Felix Divo
diff --git a/docs/examples/clustering/plot_kernel_kmeans.py b/docs/examples/clustering/plot_kernel_kmeans.py
@@ -3,19 +3,20 @@
 Kernel k-means
 ==============
 
-This example uses Global Alignment kernel (GAK, [1]) at the core of a kernel
-:math:`k`-means algorithm [2] to perform time series clustering.
+This example uses Global Alignment kernel (GAK, [1]_) at the core of a kernel
+:math:`k`-means algorithm [2]_ to perform time series clustering.
 
 Note that, contrary to :math:`k`-means, a centroid cannot be computed when
 using kernel :math:`k`-means. However, one can still report cluster
 assignments, which is what is provided here: each subfigure represents the set
 of time series from the training set that were assigned to the considered
 cluster.
 
-[1] M. Cuturi, "Fast global alignment kernels," ICML 2011.
+.. [1] M. Cuturi, "Fast global alignment kernels",
+  ICML 2011.
 
-[2] I. S. Dhillon, Y. Guan, B. Kulis. Kernel k-means, Spectral Clustering and \
-Normalized Cuts. KDD 2004.
+.. [2] I. S. Dhillon, Y. Guan, B. Kulis. Kernel k-means, Spectral Clustering and
+  Normalized Cuts. KDD 2004.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/clustering/plot_kmeans.py b/docs/examples/clustering/plot_kmeans.py
@@ -6,8 +6,8 @@
 This example uses :math:`k`-means clustering for time series. Three variants of
 the algorithm are available: standard
 Euclidean :math:`k`-means, DBA-:math:`k`-means (for DTW Barycenter
-Averaging [1])
-and Soft-DTW :math:`k`-means [2].
+Averaging [1]_)
+and Soft-DTW :math:`k`-means [2]_.
 
 In the figure below, each row corresponds to the result of a different
 clustering. In a row, each sub-figure corresponds to a cluster.
@@ -29,11 +29,11 @@
 time series is scaled independently and there is hence no such thing as an
 overall data range.
 
-[1] F. Petitjean, A. Ketterlin & P. Gancarski. A global averaging method \
-for dynamic time warping, with applications to clustering. Pattern \
-Recognition, Elsevier, 2011, Vol. 44, Num. 3, pp. 678-693
-[2] M. Cuturi, M. Blondel "Soft-DTW: a Differentiable Loss Function for \
-Time-Series," ICML 2017.
+.. [1] F. Petitjean, A. Ketterlin & P. Gancarski. A global averaging method
+  for dynamic time warping, with applications to clustering. Pattern
+  Recognition, Elsevier, 2011, Vol. 44, Num. 3, pp. 678-693
+.. [2] M. Cuturi, M. Blondel "Soft-DTW: a Differentiable Loss Function for
+  Time-Series", ICML 2017.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/clustering/plot_kshape.py b/docs/examples/clustering/plot_kshape.py
@@ -3,12 +3,12 @@
 KShape
 ======
 
-This example uses the KShape clustering method [1] that is based on
+This example uses the KShape clustering method [1]_ that is based on
 cross-correlation to cluster time series.
 
 
-[1] J. Paparrizos & L. Gravano. k-Shape: Efficient and Accurate Clustering \
-of Time Series. SIGMOD 2015. pp. 1855-1870.
+.. [1] J. Paparrizos & L. Gravano. k-Shape: Efficient and Accurate Clustering
+  of Time Series. SIGMOD 2015. pp. 1855-1870.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/metrics/plot_lb_keogh.py b/docs/examples/metrics/plot_lb_keogh.py
@@ -4,7 +4,7 @@
 ========
 
 This example illustrates the principle of time series envelope and its
-relationship to the "LB_Keogh" lower bound [1].
+relationship to the "LB_Keogh" lower bound [1]_.
 
 The envelope of a time series consists of two time series such that the
 original time series is between the two time series. Denoting the original
@@ -21,7 +21,7 @@
 
 where :math:`r` is the radius of the envelope.
 
-The distance between a time series $Q$ and an envelope :math:`(L, U)` is
+The distance between a time series :math:`Q` and an envelope :math:`(L, U)` is
 defined as:
 
 .. math::
@@ -36,8 +36,8 @@
 
 So it is simply the Euclidean distance between :math:`Q` and the envelope.
 
-[1] E. Keogh and C. A. Ratanamahatana, "Exact indexing of dynamic time
-warping". Knowledge and Information Systems, 7(3), 358-386 (2004).
+.. [1] E. Keogh and C. A. Ratanamahatana, "Exact indexing of dynamic time
+  warping". Knowledge and Information Systems, 7(3), 358-386 (2004).
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/metrics/plot_lcss.py b/docs/examples/metrics/plot_lcss.py
@@ -4,20 +4,20 @@
 ==========================
 
 This example illustrates LCSS computation between time series and plots the
-alignment path [1]. and its relationship to the DTW.
+alignment path [1]_. and its relationship to the DTW.
 
 Since LCSS focuses on the similar parts between two time-series, a potential
 use case is to identify the similarity between time-series whose lengths differ
-greatly or have noise. As one example, M. Vlachos et al. [1] used this method
+greatly or have noise. As one example, M. Vlachos et al. [1]_ used this method
 to cluster time series regarding human writing in the presence of noise.
 
 The example demonstrates the use of the functions `lcss_path` and `dtw_path`
 to calculate the alignment path between them and compare the two approaches
 when dealing with unequal-length sequence data and noise.
 
-[1] M. Vlachos, D. Gunopoulos, and G. Kollios. 2002. "Discovering Similar
-Multidimensional Trajectories", In Proceedings of the 18th International
-Conference on Data Engineering (ICDE '02). IEEE Computer Society, USA, 673.
+.. [1] M. Vlachos, D. Gunopoulos, and G. Kollios. 2002. "Discovering Similar
+  Multidimensional Trajectories", In Proceedings of the 18th International
+  Conference on Data Engineering (ICDE '02). IEEE Computer Society, USA, 673.
 """
 
 # Author: Daniela Duarte
diff --git a/docs/examples/metrics/plot_sdtw.py b/docs/examples/metrics/plot_sdtw.py
@@ -7,7 +7,7 @@
 matches of a sequence in a longer sequence.
 
 A potential usecase is to identify the occurrence of certain events in
-continuous sensor signals. As one example Barth et al. [1] used this method
+continuous sensor signals. As one example Barth et al. [1]_ used this method
 to find stride in sensor recordings of gait.
 
 The example demonstrates the use of the functions
@@ -16,9 +16,9 @@
 you are only interested in finding the optimal alignment, you can directly use
 `dtw_subsequence_path`.
 
-[1] Barth, et al. (2013): Subsequence dynamic time warping as a method for \
-robust step segmentation using gyroscope signals of daily life activities, \
-EMBS, https://doi.org/10.1109/EMBC.2013.6611104
+.. [1] Barth, et al. (2013): Subsequence dynamic time warping as a method for \
+  robust step segmentation using gyroscope signals of daily life activities, \
+  EMBS, https://doi.org/10.1109/EMBC.2013.6611104
 
 """
 
diff --git a/docs/examples/misc/plot_matrix_profile.py b/docs/examples/misc/plot_matrix_profile.py
@@ -3,16 +3,16 @@
 Matrix Profile
 ==============
 
-This example presents a toy example of using Matrix Profile [1] for anomaly 
+This example presents a toy example of using Matrix Profile [1]_ for anomaly
 detection.
 
 Matrix Profile transforms a time series into a sequence of 1-Nearest-Neighbor 
 distances between its subseries.
 
-[1] C. M. Yeh, Y. Zhu, L. Ulanova, N.Begum et al. 
-Matrix Profile I: All Pairs Similarity Joins for Time Series: A 
-Unifying View that Includes Motifs, Discords and Shapelets.
-ICDM 2016.
+.. [1] C. M. Yeh, Y. Zhu, L. Ulanova, N.Begum et al.
+  Matrix Profile I: All Pairs Similarity Joins for Time Series: A
+  Unifying View that Includes Motifs, Discords and Shapelets.
+  ICDM 2016.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/examples/misc/plot_sax.py b/docs/examples/misc/plot_sax.py
@@ -3,7 +3,7 @@
 PAA and SAX features
 ====================
 
-This example presents a comparison between PAA [1], SAX [2] and 1d-SAX [3]
+This example presents a comparison between PAA [1]_, SAX [2]_ and 1d-SAX [3]_
 features.
 
 PAA (Piecewise Aggregate Approximation) corresponds to a downsampling of the
@@ -18,15 +18,15 @@
 by an affine function (2 parameters per segment are hence quantized: slope and
 mean value).
 
-[1] E. Keogh & M. Pazzani. Scaling up dynamic time warping for datamining
-applications. SIGKDD 2000, pp. 285--289.
+.. [1] E. Keogh & M. Pazzani. Scaling up dynamic time warping for datamining
+  applications. SIGKDD 2000, pp. 285--289.
 
-[2] J. Lin, E. Keogh, L. Wei, et al. Experiencing SAX: a novel symbolic
-representation of time series. Data Mining and Knowledge Discovery,
-2007. vol. 15(107)
+.. [2] J. Lin, E. Keogh, L. Wei, et al. Experiencing SAX: a novel symbolic
+  representation of time series. Data Mining and Knowledge Discovery,
+  2007. vol. 15(107)
 
-[3] S. Malinowski, T. Guyet, R. Quiniou, R. Tavenard. 1d-SAX: a Novel
-Symbolic Representation for Time Series. IDA 2013.
+.. [3] S. Malinowski, T. Guyet, R. Quiniou, R. Tavenard. 1d-SAX: a Novel
+  Symbolic Representation for Time Series. IDA 2013.
 """
 
 # Author: Romain Tavenard
diff --git a/docs/integration_other_software.rst b/docs/integration_other_software.rst
@@ -13,7 +13,7 @@ to convert data sets from and to other formats.
 transforms an array-like object into a three-dimensional array of shape
 ``(n_ts, sz, d)`` with the following conventions:
 
-- the fist axis is the sample axis, ``n_ts`` being the number of time series;
+- the first axis is the sample axis, ``n_ts`` being the number of time series;
 - the second axis is the time axis, ``sz`` being the maximum number of time points;
 - the third axis is the dimension axis, ``d`` being the number of dimensions.
 
diff --git a/docs/variablelength.rst b/docs/variablelength.rst
@@ -1,7 +1,7 @@
 Methods for variable-length time series
 =======================================
 
-This page lists machine learning methods in `tslearn` that are able to deal
+This page lists machine learning methods in ``tslearn`` that are able to deal
 with datasets containing time series of different lengths.
 We also provide example usage for these methods using the following
 variable-length time series dataset:
@@ -76,6 +76,7 @@ Clustering
 
 * :class:`tslearn.clustering.KernelKMeans`
 * :class:`tslearn.clustering.TimeSeriesKMeans`
+* :class:`tslearn.clustering.TimeSeriesDBSCAN`
 * :class:`tslearn.clustering.silhouette_score`
 
 Examples
@@ -87,6 +88,12 @@ Examples
     gak_km = KernelKMeans(n_clusters=2, kernel="gak")
     labels_gak = gak_km.fit_predict(X)
 
+.. code-block:: python
+
+    from tslearn.clustering import TimeSeriesDBSCAN
+    dbscan = TimeSeriesDBSCAN(min_ts=2, eps=1, metric="dtw")
+    labels_dbscan = dbscan.fit_predict(X)
+
 .. code-block:: python
 
     from tslearn.clustering import TimeSeriesKMeans
@@ -107,7 +114,6 @@ Examples
 Barycenter computation
 ----------------------
 
-
 * :class:`tslearn.barycenters.dtw_barycenter_averaging`
 * :class:`tslearn.barycenters.softdtw_barycenter`
 
@@ -126,6 +132,27 @@ Examples
     initial_barycenter = ts_zeros(sz=5)
     bar = softdtw_barycenter(X, init=initial_barycenter)
 
+Preprocessing
+-------------
+
+* :class:`tslearn.preprocessing.TimeSeriesScalerMinMax`
+* :class:`tslearn.preprocessing.TimeSeriesScalerMeanVariance`
+* :class:`tslearn.preprocessing.TimeSeriesImputer`
+
+Examples
+~~~~~~~~
+
+.. code-block:: python
+
+    from tslearn.preprocessing import (
+        TimeSeriesScalerMinMax,
+        TimeSeriesScalerMeanVariance,
+        TimeSeriesImputer
+    )
+    scaled = TimeSeriesScalerMinMax().fit_transform(X)
+    scaled = TimeSeriesScalerMeanVariance().fit_transform(X)
+    imputed = TimeSeriesImputer().fit_transform(X)
+
 Model selection
 ---------------
 
diff --git a/tslearn/barycenters/softdtw.py b/tslearn/barycenters/softdtw.py
@@ -37,7 +37,7 @@ def _softdtw_func(Z, X, weights, barycenter, gamma):
 
 def softdtw_barycenter(X, gamma=1.0, weights=None, method="L-BFGS-B", tol=1e-3,
                        max_iter=50, init=None):
-    """Compute barycenter (time series averaging) under the soft-DTW [1]
+    """Compute barycenter (time series averaging) under the soft-DTW
     geometry.
 
     Soft-DTW was originally presented in [1]_.
diff --git a/tslearn/neighbors/neighbors.py b/tslearn/neighbors/neighbors.py
diff --git a/tslearn/preprocessing/preprocessing.py b/tslearn/preprocessing/preprocessing.py
