Choosing the number of clusters in heirarchical agglomerative clustering with scikit - machine-learning

The wikipedia article on determining the number of clusters in a dataset indicated that I do not need to worry about such a problem when using hierarchical clustering. However when I tried to use scikit-learn's agglomerative clustering I see that I have to feed it the number of clusters as a parameter "n_clusters" - without which I get the hardcoded default of two clusters. How can I go about choosing the right number of cluster's for the dataset in this case? Is the wiki article wrong?

Wikipedia is simply making an extreme simplification which has nothing to do with real life. Hierarchical clustering does not avoid the problem with number of clusters. Simply - it constructs the tree spaning over all samples, which shows which samples (later on - clusters) merge together to create a bigger cluster. This happend recursively till you have just two clusters (this is why default number of clusters is 2) which are merged to the whole dataset. You are left alone with "cutting" through the tree to get actual clustering. Once you fit AgglomerativeClustering you can traverse the whole tree and analyze which clusters to keep
import numpy as np
from sklearn.cluster import AgglomerativeClustering
import itertools
X = np.concatenate([np.random.randn(3, 10), np.random.randn(2, 10) + 100])
clustering = AgglomerativeClustering()
clustering.fit(X)
[{'node_id': next(itertools.count(X.shape[0])), 'left': x[0], 'right':x[1]} for x in clustering.children_]

ELKI (not scikit-learn, but Java) has a number of advanced methods that extract clusters from a hierarchical clustering. They are smarter than just cutting the tree at a particular height, but they can produce a hierarchy of clusters of a minimum size, for example.
You could check if these methods work for you.

Related

Is there a way to find the most representative set of samples of the entire dataset?

I'm working on text classification and I have a set of 200.000 tweets.
The idea is to manually label a short set of tweets and train classifiers to predict the labels of the rest. Supervised learning.
What I would like to know is if there is a method to choose what samples to include in the train set in a way that this train set is a good representation of the whole data set, and because the high diversity included in the train set, the trained classifiers have considerable trust to be applied on the rest of tweets.
This sounds like a stratification question - do you have pre-existing labels or do you plan to design the labels based on the sample you're constructing?
If it's the first scenario, I think the steps in order of importance would be:
Stratify by target class proportions (so if you have three classes, and they are 50-30-20%, train/dev/test should follow the same proportions)
Stratify by features you plan to use
Stratify by tweet length/vocabulary etc.
If it's the second scenario, and you don't have labels yet, you may want to look into using n-grams as a feature, coupled with a dimensionality reduction or clustering approach. For example:
Use something like PCA or t-SNE to maximize distance between tweets (or a large subset), then pick candidates from different regions of the projected space
Cluster them based on lexical items (unigrams or bigrams, possibly using log frequencies or TF-IDF and stop word filtering, if content words are what you're looking for) - then you can cut the tree at a height that gives you n bins, which you can then use as a source for samples (stratify by branch)
Use something like LDA to find n topics, then sample stratified by topic
Hope this helps!
It seems that before you know anything about the classes you are going to label, a simple uniform random sample will do almost as well as any stratified sample - because you don't know in advance what to stratify on.
After labelling this first sample and building the first classifier, you can start so-called active learning: make predictions for the unlabelled dataset, and sample some tweets in which your classifier is least condfident. Label them, retrain the classifier, and repeat.
Using this approach, I managed to create a good training set after several (~5) iterations, with ~100 texts in each iteration.

Hierarchical Clustering

I have read some resources and I found out how hierarchical clustering works. However, when I compare it with k-means clustering, it seems to me that k-means really constitues specific number of clusters,whereas hierarchical analysis shows me how the samples can be clustered. What I mean is that I do not get a specific number of clusters in hierarchical clustering. I get only a scheme about how the clusters can be constituted and portion of relation between the samples.
Thus, I cannot understand where I can use this clustering method.
Hierarchical clustering (HC) is just another distance-based clustering method like k-means. The number of clusters can be roughly determined by cutting the dendrogram represented by HC. Determining the number of clusters in a data set is not an easy task for all clustering methods, which is usually based on your applications. Tuning the thresholds in HC may be more explicit and straightforward for researchers, especially for a very large data set. I think this question is also related.
In k-means clustering k is a hyperparameter that you need to find in order to divide your data points into clusters whereas in hierarchical clustering (lets take one type of hierarchical clustering i.e. agglomerative) firstly you consider all the points in your dataset as a cluster and then merge two clusters based on a similarity metric and repeat this until you get a single cluster. I will explain this with an example.
Suppose initially you have 13 points (x_1,x_2,....,x_13) in your dataset so at start you will have 13 clusters, now in second step lets you get 7 clusters (x_1-x_2 , x_4-x_5, x_6-x_8, x_3-x_7, x_11-x_12, x_10, x_13) based on the similarity between the points. In the third step lets say you get 4 clusters(x_1-x_2-x_4-x_5, x_6-x_8-x_10, x_3-x_7-x_13, x_11-x_12) like this you would arrive to a step wherein all the points in your dataset form one cluster and which is also the last step of agglomerative clustering algorithm.
So in hierarchical clustering, there is no hyperparameter, depending upon your problem, if you want 7 clusters then stop at the second step if you want 4 clusters then stop at the third step and likewise.
A practical advantage in hierarchical clustering is the possibility of visualizing results using dendrogram. If you don’t know in advance what number of clusters you’re looking for (as is often the case…), you can use the dendrogram plot that can help you choose k with no need to create separate clusterings. Dendrogram can also give a great insight into the data structure, help identify outliers, etc. Hierarchical clustering is also deterministic, whereas k-means with random initialization can give you different results when running several times on the same data.
Hope this helps.

Iterative modularity

I have run modulartiy edge_weight/randomized at a resolution of 1, atleast 20 times on the same network. This is the same network I have created based on the following rule. Two nodes are related if they have atleast one item in common. Every time I run modularity I get a little different node distribution among communities. Additionally, I get 9 or 10 communities but it is never consistent. Any comment or help is much appreciated.
I found a solution to my problem using consensus clustering. Here is the paper that describes it. One way to get the optimum clusters without having to solve them in a high-dimensional space using spectral clustering would be to run the algorithm repeatedly until no more partitions can be achieved. Here is the article and complete explanation details:
SCIENTIFIC REPORTS | 2 : 336 | DOI: 10.1038/srep00336
Consensus clustering in complex networks Andrea Lancichinetti & Santo Fortunato
The consensus matrix. Let us suppose that we wish to combine nP partitions found by a clustering algorithm on a network with n vertices. The consensus matrix D is an n x n matrix, whose entry Dij indicates the number of partitions in which vertices i and j of the network were assigned to the same cluster, divided by the number of partitions nP. The matrix D is usually much denser than the adjacency matrix A of the original network, because in the consensus matrix there is an edge between any two vertices which have cooccurred in the same cluster at least once. On the other hand, the weights are large only for those vertices which are most frequently coclustered, whereas low weights indicate that the vertices are probably at the boundary between different (real) clusters, so their classification in the same cluster is unlikely and essentially due to noise. We wish to maintain the large weights and to drop the low ones, therefore a filtering procedure is in order. Among the other things, in the absence of filtering the consensus matrix would quickly grow into a very dense matrix, which would make the application of any clustering algorithm computationally expensive.We discard all entries of D below a threshold t. We stress that there might be some noisy vertices whose edges could all be below the threshold, and they would be not connected anymore. When this happens, we just connect them to their neighbors with highest weights, to keep the graph connected all along the procedure.
Next we apply the same clustering algorithm to D and produce another set of partitions, which is then used to construct a new consensus matrix D9, as described above. The procedure is iterated until the consensus matrix turns into a block diagonal matrix Dfinal, whose weights equal 1 for vertices in the same block and 0 for vertices in different blocks. The matrix Dfinal delivers the community structure of the original network. In our calculations typically one iteration is sufficient to lead to stable results. We remark that in order to use the same clustering method all along, the latter has to be able to detect clusters in weighted networks, since the consensus matrix is weighted. This is a necessary constraint on the choice of the methods for which one could use the procedure proposed here. However, it is not a severe limitation,as most clustering algorithms in the literature can handle weighted networks or can be trivially extended to deal with them.
I think that the answer is in the randomizing part of the algorithm. You can find more details here:
https://github.com/gephi/gephi/wiki/Modularity
https://sites.google.com/site/findcommunities/
http://lanl.arxiv.org/abs/0803.0476

Clustering Method Selection in High-Dimension?

If the data to cluster are literally points (either 2D (x, y) or 3D (x, y,z)), it would be quite intuitive to choose a clustering method. Because we can draw them and visualize them, we somewhat know better which clustering method is more suitable.
e.g.1 If my 2D data set is of the formation shown in the right top corner, I would know that K-means may not be a wise choice here, whereas DBSCAN seems like a better idea.
However, just as the scikit-learn website states:
While these examples give some intuition about the algorithms, this
intuition might not apply to very high dimensional data.
AFAIK, in most of the piratical problems we don't have such simple data. Most probably, we have high-dimensional tuples, which cannot be visualized like such, as data.
e.g.2 I wish to cluster a data set where each data is represented as a 4-D tuple <characteristic1, characteristic2, characteristic3, characteristic4>. I CANNOT visualize it in a coordinate system and observes its distribution like before. So I will NOT be able to say DBSCAN is superior to K-means in this case.
So my question:
How does one choose the suitable clustering method for such an "invisualizable" high-dimensional case?
"High-dimensional" in clustering probably starts at some 10-20 dimensions in dense data, and 1000+ dimensions in sparse data (e.g. text).
4 dimensions are not much of a problem, and can still be visualized; for example by using multiple 2d projections (or even 3d, using rotation); or using parallel coordinates. Here's a visualization of the 4-dimensional "iris" data set using a scatter plot matrix.
However, the first thing you still should do is spend a lot of time on preprocessing, and finding an appropriate distance function.
If you really need methods for high-dimensional data, have a look at subspace clustering and correlation clustering, e.g.
Kriegel, Hans-Peter, Peer Kröger, and Arthur Zimek. Clustering high-dimensional data: A survey on subspace clustering, pattern-based clustering, and correlation clustering. ACM Transactions on Knowledge Discovery from Data (TKDD) 3.1 (2009): 1.
The authors of that survey also publish a software framework which has a lot of these advanced clustering methods (not just k-means, but e.h. CASH, FourC, ERiC): ELKI
There are at least two common, generic approaches:
One can use some dimensionality reduction technique in order to actually visualize the high dimensional data, there are dozens of popular solutions including (but not limited to):
PCA - principal component analysis
SOM - self-organizing maps
Sammon's mapping
Autoencoder Neural Networks
KPCA - kernel principal component analysis
Isomap
After this one goes back to the original space and use some techniques that seems resonable based on observations in the reduced space, or performs clustering in the reduced space itself.First approach uses all avaliable information, but can be invalid due to differences induced by the reduction process. While the second one ensures that your observations and choice is valid (as you reduce your problem to the nice, 2d/3d one) but it loses lots of information due to transformation used.
One tries many different algorithms and choose the one with the best metrics (there have been many clustering evaluation metrics proposed). This is computationally expensive approach, but has a lower bias (as reducting the dimensionality introduces the information change following from the used transformation)
It is true that high dimensional data cannot be easily visualized in an euclidean high dimensional data but it is not true that there are no visualization techniques for them.
In addition to this claim I will add that with just 4 features (your dimensions) you can easily try the parallel coordinates visualization method. Or simply try a multivariate data analysis taking two features at a time (so 6 times in total) to try to figure out which relations intercour between the two (correlation and dependency generally). Or you can even use a 3d space for three at a time.
Then, how to get some info from these visualizations? Well, it is not as easy as in an euclidean space but the point is to spot visually if the data clusters in some groups (eg near some values on an axis for a parallel coordinate diagram) and think if the data is somehow separable (eg if it forms regions like circles or line separable in the scatter plots).
A little digression: the diagram you posted is not indicative of the power or capabilities of each algorithm given some particular data distributions, it simply highlights the nature of some algorithms: for instance k-means is able to separate only convex and ellipsoidail areas (and keep in mind that convexity and ellipsoids exist even in N-th dimensions). What I mean is that there is not a rule that says: given the distributiuons depicted in this diagram, you have to choose the correct clustering algorithm consequently.
I suggest to use a data mining toolbox that lets you explore and visualize the data (and easily transform them since you can change their topology with transformations, projections and reductions, check the other answer by lejlot for that) like Weka (plus you do not have to implement all the algorithms by yourself.
In the end I will point you to this resource for different cluster goodness and fitness measures so you can compare the results rfom different algorithms.
I would also suggest soft subspace clustering, a pretty common approach nowadays, where feature weights are added to find the most relevant features. You can use these weights to increase performance and improve the BMU calculation with euclidean distance, for example.

Clustering Baseline Comparison, KMeans

I'm working on an algorithm that makes a guess at K for kmeans clustering. I guess I'm looking for a data set that I could use as a comparison, or maybe a few data sets where the number of clusters is "known" so I could see how my algorithm is doing at guessing K.
I would first check the UCI repository for data sets:
http://archive.ics.uci.edu/ml/datasets.html?format=&task=clu&att=&area=&numAtt=&numIns=&type=&sort=nameUp&view=table
I believe there are some in there with the labels.
There are text clustering data sets that are frequently used in papers as baselines, such as 20newsgroups:
http://qwone.com/~jason/20Newsgroups/
Another great method (one that my thesis chair always advocated) is to construct your own small example data set. The best way to go about this is to start small, try something with only two or three variables that you can represent graphically, and then label the clusters yourself.
The added benefit of a small, homebrew data set is that you know the answers and it is great for debugging.
Since you are focused on k-means, have you considered using the various measures (Silhouette, Davies–Bouldin etc.) to find the optimal k?
In reality, the "optimal" k may not be a good choice. Most often, one does want to choose a much larger k, then analyze the resulting clusters / prototypes in more detail to build clusters out of multiple k-means partitions.
The iris flower dataset is a good one to start with, that clustering works nicely on.
Download here

Resources