We Keep Coding

Clustering on evenly spaced grid points

I have a 50 by 50 grid of evenly spaced (x,y) points. Each of these points has a third scalar value. This can be visualized using a contourplot which I have added. I am interested in the regions indicated in by the red circles. These regions of low "Z-values" are what I want to extract from this data.
2D contour plot of 50 x 50 evenly spaced grid points:
I want to do this by using clustering (machine learning), which can be lightning quick when applied correctly. The problem is, however, that the points are evenly spaced together and therefore the density of the entire dataset is equal everywhere.
I have tried using a DBSCAN algorithm with a custom distance metric which takes into account the Z values of each point. I have defined the distance between two points as follows:\
def custom_distance(point1,point2):
average_Z = (point1[2]+point2[2])/2
distance = np.sqrt(np.square((point1[0]-point2[0])) + np.square((point1[1]-point2[1])))
distance = distance * average_Z
return distance
This essentially determines the Euclidean distance between two points and adds to it the average of the two Z values of both points. In the picture below I have tested this distance determination function applied in a DBSCAN algorithm. Each point in this 50 by 50 grid each has a Z value of 1, except for four clusters that I have randomly placed. These points each have a z value of 10. The algorithm is able to find the clusters in the data based on their z value as can be seen below.
DBSCAN clustering result using scalar value distance determination:
Positive about the results I tried to apply it to my actual data, only to be disappointed by the results. Since the x and y values of my data are very large, I have simply scaled them to be 0 to 49. The z values I have left untouched. The results of the clustering can be seen in the image below:
Clustering result on original data:
This does not come close to what I want and what I was expecting. For some reason the clusters that are found are of rectangular shape and the light regions of low Z values that I am interested in are not extracted with this approach.
Is there any way I can make the DBSCAN algorithm work in this way? I suspect the reason that it is currently not working has something to do with the differences in scale of the x,y and z values. I am also open for tips or recommendations on other approaches on how to define and find the lighter regions in the data.

Measuring dissimilarity within the cluster - Kmodes

I was doing clustering with categorical data. I came across Kmodes algo and found it to be perfect for my requirements. Now, I want to measure dissimilarity within a cluster for all the clusters. I am thinking to measure the dissimilarity with a cluster and reduce it as much as possible. Is there any way to do that?
Alternatively, is there any way to check how efficiently my data has been clustered?
Since my data is categorical, ways which consider distance as a metric might not be helpful.

To measure the dissimilarity within a cluster you need to come up with some kind of a metric. For categorical data, one of the possible ways of calculating dissimilarity could be the following:
d(i, j) = (p - m) / p
where:
p is the number of classes/categories in your data
m is the number of matches you have between samples i and j
For example, if your data has 3 categorical features and the samples, i and j are as follows:
Feature1 Feature2 Feature3
i x y z
j x w z
So here, we have 3 categorical features, so p=3 and out of these three, two features have same values for the samples i and j, so m=2. Therefore
d(i,j) = (3 - 2) / 3
d(i,j) = 0.33
Another alternative is to convert your categorical variables to one-hot-encoded features and then compute jaccard simmilarity.
So, in order to measure the dissimilarity within a cluster you could calculate pairwise dissimilarity between each object in your cluster and then take the average of that.
Based on these measures you may also use the silhoutte score for evaluating the quality of your clustering (but you need to take it with a grain of salt, sometimes the score can be good while the clustering might not be what you expected).

Evaluating vector distance measures

I am working with vectors of word frequencies and trying out some of the different distance measures available in Scikit Learns Pairwise Distances. I would like to use these distances for clustering and classification.
I usually have a feature matrix of ~ 30,000 x 100. My idea was to choose a distance metric that maximizes the pairwise distances by running pairwise differences over the same dataset with the distance metrics available in Scipy (e.g. Euclidean, Cityblock, etc.) and for each metric
convert distances computed for the dataset to zscores to normalize across metrics
get the range of these zscores, i.e. the spread of the distances
use the distance metric that gives me the widest range of distances as it apparently gives me the maximum spread over my dataset and the most variance to work with. (Cf. code below)
My questions:
Does this approach make sense?
Are there other evaluation procedures that one should try? I found these papers (Gavin, Aggarwal, but they don't apply 100 % here...)
Any help is much appreciated!
My code:
matrix=np.random.uniform(0, .1, size=(10,300)) #test data set
scipy_distances=['euclidean', 'minkowski', ...] #these are the distance metrics
for d in scipy_distances: #iterate over distances
distmatrix=sklearn.metrics.pairwise.pairwise_distances(matrix, metric=d)
distzscores = scipy.stats.mstats.zscore(distmatrix, axis=0, ddof=1)
diststats=basicstatsmaker(distzscores)
range=np.ptp(distzscores, axis=0)
print "range of metric", d, np.ptp(range)

In general - this is just a heuristic, which might, or not - work. In particular, it is easy to construct a "dummy metric" which will "win" in your approach even though it is useless. Try out
class Dummy_dist:
def __init__(self):
self.cheat = True
def __call__(self, x, y):
if self.cheat:
self.cheat = False
return 1e60
else:
return 0
dummy_dist = Dummy_dist()
This will give you huuuuge spread (even with z-score normalization). Of course this is a cheating example as this is non determinsitic, but I wanted to show the basic counterexample, and of course given your data one can construct a deterministic analogon.
So what you should do? Your metric should be treated as hyperparameter of your process. You should not divide process of generating your clustering/classification into two separate phases: choosing a distance and then learning something; but you should do this jointly, consider your clustering/classification + distance pairs as a single model, thus instead of working with k-means, you will work with k-means+euclidean, k-means+minkowsky and so on. This is the only statistically supported approach. You cannot construct a method of assessing "general goodness" of the metric, as there is no such object, metric quality can be only assessed in a particular task, which involves fixing every other element (such as a clustering/classification method, particular dataset etc.). Once you perform such wide, exhaustive evaluation, check many such pairs, on many datasets, you might claim that given metric performes best in such range of tasks.

Python: how to compare the similarity between clustering using k-means algorithm?

I have two observations of the same event. Let say X and Y.
I suppose to have nc clusters. I am using sklearn to make the clustering.
x = KMeans(n_clusters=nc).fit_predict(X)
y = KMeans(n_clusters=nc).fit_predict(Y)
is there a measure that allow me to compare x and y: i.e. this measure will be 1 if the clusters x and y are the same.

Just extract the cluster centers of your kmeans-objects (see the docs):
x_centers = x.cluster_centers_
y_centers = y.cluster_centers_
The you have to decide which metric you are using to compare these. Keep in mind that the centers are floating-points, the clustering-process is a heuristic and the clustering-process is a random-algorithm. This means, you will get something which interprets as not exactly the same with a high probability, even for cluster-objects trained on the same data.
This link discusses some approaches and the problems.

The Rand Index and its adjusted version do this exactly. Two cluster assignments that match (even if the labels themselves, which are treated as arbitrary, are different), get a score of 1. A value of 0 means they don't agree at all. The Adjusted Rand Index uses its baseline as random assignment of points to clusters.

Use K-means to learn features in Python

Question
I implemented a K-Means algorithm in Python. First I apply PCA and whitening to the input data. Then I use k-means to successfully subtract k centroids out of the data.
How can I use those centroids to understand the "features" learnt? Are the centroids already the features (doesn't seem like this to me) or do I need to combine them with the input data again?
Because of some answers: K-means is not "just" a method for clustering, instead it's a vector quantization method. That said the goal of k-means is to describe a dataset with a reduced number of feature vectors. Therefore there are big analogies to methods like Sparse Filtering/ Learning regarding the potential outcome.
Code Example
# Perform K-means, data already pre-processed
centroids = k_means(matrix_pca_whitened,1000)
# Assign data to centroid
idx,_ = vq(song_matrix_pca,centroids)

The clusters produced by the K-mean algorithms separate your input space into K regions. When you have new data, you can tell which region it belongs to, and thus classify it.
The centroids are just a property of these clusters.
You can have a look at the scikit-learn doc if you are unsure, and at the map to make sure you choose the right algorithm.

This is sort of a circular question: "understand" requires knowing something about the features outside of the k-means process. All that k-means does is to identify k groups of physical proximity. It says "there are clumps of stuff in these 'k' places, and here's how the all the points choose the nearest."
What this means in terms of the features is up to the data scientist, rather than any deeper meaning that k-means can ascribe. The variance of each group may tell you a little about how tightly those points are clustered. Do remember that k-means also chooses starting points at random; an unfortunate choice can easily give a sub-optimal description of the space.
A centroid is basically the "mean" of the cluster. If you can ascribe some deeper understanding from the distribution of centroids, great -- but that depends on the data and features, rather than any significant meaning devolving from k-means.
Is that the level of answer you need?

The centroids are in fact the features learnt. Since k-means is a method of vector quantization we look up which observation belongs to which cluster and therefore is best described by the feature vector (centroid).
By having one observation e.g. separated into 10 patches before, the observation might consist of 10 feature vectors max.
Example:
Method: K-means with k=10
Dataset: 20 observations divided into 2 patches each = 40 data vectors
We now perform K-means on this patched dataset and get the nearest centroid per patch. We could then create a vector for each of the 20 observations with the length 10 (=k) and if patch 1 belongs to centroid 5 and patch 2 belongs to centroid 9 the vector could look like: 0 - 0 - 0 - 0 - 1 - 0 - 0 - 0 - 1 - 0.
This means that this observation consists of the centroids/ features 5 and 9. You could also measure use the distance between patch and centroid instead of this hard assignment.