New answers tagged

0

NMF provides weights. I’m not sure if these weights can be normalized and considered to be probabilities, but they can be used in a similar manner to identify more and less likely/weighted. The question would be if LDA is well-calibrated and if NMF weights could be used in a manner that ranks things in a well-calibrated way. How do you intend to use the ...


1

DTW does not have a notion of "convergence", as it is not an iterative optimization procedure. When DTW is too slow, and doesn't finish in acceptable time, the obviously next best idea is to use bounded DTW. Full DTW can take O(nm) time of n and m are the lengths of your series (at least if implemented well...). You can bound this to O(nk) for a fixed ...


2

You question is almost meaningless without showing/explaining the data. The tells you want invarinces you need (phase. offset, amplitude, uniform scaling, warping, complexity etc [a]). But see for example [b] [a] https://www.cs.ucr.edu/~eamonn/Complexity-Invariant%20Distance%20Measure.pdf [b] https://www.cs.ucr.edu/~eamonn/time_series_bitmaps.pdf


1

n is not related to the class size. It's a sample size you need to choose according to how precise you want the estimate to be. A larger sample will give you a better estimate. IMHO, Hopkins is useless for testing clusterability. It is a test for multivariate uniform distributions. But that never is a problem that your data is uniform, is it? To see the ...


1

This is roughly the idea of bisecting k-means. No, the result is almost certainly worse (by SSQ) than that of k-means with the same number of clusters. K-means finds (at least almost) a local optimum. The bisection approach does not, as such optimas are usually not hierarchic (the best k=2 and the best k=3 solution do usually not have a center in common, ...


2

Sure, it could - but only if your dataset has a very specific type of structure. And there's no reason to expect that specific structure to be a common feature of datasets, so I don't expect this would be a generally useful process. That doesn't mean that k-means is great by itself; there are many modified versions of the algorithm that have proven to be ...


1

There is plenty of literature on unsupervised outlier detection. For example the Local Outlier Factor (Wikipedia), and a simple k-nearest-neighbor-distance based approaches seem to work for many data sets.


0

It's one of her many combinations that does not make a lot of sense when you bother to try to understand the methods. SVD tried to generate a projection where there is some kind of linearity in the factors. KernelPCA is used when you know non-linear relationships exist in your data. It's generally a bad idea to just stack method on method that you don't ...


1

You can use OneClassSVM or IsolationForest. The purpose of these algorithms are to find what is normal by learning the distribution of the normal things. You need to fit them with normal data, then they will do a binary classification and say what is not normal. This is also the idea behind auto encoder in anomaly detection (you point out in your article). ...


Top 50 recent answers are included