There have recently been significant advances in the accuracy of algorithms proposed for time series classification (TSC). However, a commonly asked question by real world practitioners and data scientists less familiar with the research topic, is whether the complexity of the algorithms considered state of the art is really necessary. Many times the first approach suggested is a simple pipeline of summary statistics or other time series feature extraction approaches such as TSFresh, which in itself is a sensible question; in publications on TSC algorithms generalised for multiple problem types, we rarely see these approaches considered or compared against. We experiment with basic feature extractors using vector based classifiers shown to be effective with continuous attributes in current state-of-the-art time series classifiers. We test these approaches on the UCR time series dataset archive, looking to see if TSC literature has overlooked the effectiveness of these approaches. We find that a pipeline of TSFresh followed by a rotation forest classifier, which we name FreshPRINCE, performs best. It is not state of the art, but it is significantly more accurate than nearest neighbour with dynamic time warping, and represents a reasonable benchmark for future comparison. Keywords: Time series classification; transformation based classification; time series pipeline.

We assess whether using six smoothing algorithms (moving average, exponential smoothing, Gaussian filter, Savitzky-Golay filter, Fourier approximation and a recursive median sieve) could be automatically applied to time series classification problems as a preprocessing step to improve the performance of three benchmark classifiers (1-Nearest Neighbour with Euclidean and Dynamic Time Warping distances, and Rotation Forest). We found no significant improvement over unsmoothed data even when we set the smoothing parameter through cross validation. We are not claiming smoothing has no worth. It has an important role in exploratory analysis and helps with specific classification problems where domain knowledge can be exploited. What we observe is that the automatic application does not help and that we cannot explain the improvement of other time series classification algorithms over the baseline classifiers simply as a function of the absence of smoothing.

Rotation forest is a tree based ensemble that performs transforms on subsets of attributes prior to constructing each tree. We present an empirical comparison of classifiers for problems with only real valued features. We evaluate classifiers from three families of algorithms: support vector machines; tree-based ensembles; and neural networks. We compare classifiers on unseen data based on the quality of the decision rule (using classification error) the ability to rank cases (area under the receiver operator curve) and the probability estimates (using negative log likelihood). We conclude that, in answer to the question posed in the title, yes, rotation forest, is significantly more accurate on average than competing techniques when compared on three distinct sets of datasets. The same pattern of results are observed when tuning classifiers on the train data using a grid search. We investigate why rotation forest does so well by testing whether the characteristics of the data can be used to differentiate classifier performance. We assess the impact of the design features of rotation forest through an ablative study that transforms random forest into rotation forest. We identify the major limitation of rotation forest as its scalability, particularly in number of attributes. To overcome this problem we develop a model to predict the train time of the algorithm and hence propose a contract version of rotation forest where a run time cap {\em a priori}. We demonstrate that on large problems rotation forest can be made an order of magnitude faster without significant loss of accuracy and that there is no real benefit (on average) from tuning the ensemble. We conclude that without any domain knowledge to indicate an algorithm preference, rotation forest should be the default algorithm of choice for problems with continuous attributes.

We introduce canonical correlation forests (CCFs), a new decision tree ensemble method for classification and regression. Individual canonical correlation trees are binary decision trees with hyperplane splits based on local canonical correlation coefficients calculated during training. Unlike axis-aligned alternatives, the decision surfaces of CCFs are not restricted to the coordinate system of the inputs features and therefore more naturally represent data with correlated inputs. CCFs naturally accommodate multiple outputs, provide a similar computational complexity to random forests, and inherit their impressive robustness to the choice of input parameters. As part of the CCF training algorithm, we also introduce projection bootstrapping, a novel alternative to bagging for oblique decision tree ensembles which maintains use of the full dataset in selecting split points, often leading to improvements in predictive accuracy. Our experiments show that, even without parameter tuning, CCFs out-perform axis-aligned random forests and other state-of-the-art tree ensemble methods on both classification and regression problems, delivering both improved predictive accuracy and faster training times. We further show that they outperform all of the 179 classifiers considered in a recent extensive survey.

We demonstrate that, for a range of state-of-the-art machine learning algorithms, the differences in generalisation performance obtained using default parameter settings and using parameters tuned via cross-validation can be similar in magnitude to the differences in performance observed between state-of-the-art and uncompetitive learning systems. This means that fair and rigorous evaluation of new learning algorithms requires performance comparison against benchmark methods with best-practice model selection procedures, rather than using default parameter settings. We investigate the sensitivity of three key machine learning algorithms (support vector machine, random forest and rotation forest) to their default parameter settings, and provide guidance on determining sensible default parameter values for implementations of these algorithms. We also conduct an experimental comparison of these three algorithms on 121 classification problems and find that, perhaps surprisingly, rotation forest is significantly more accurate on average than both random forest and a support vector machine.