Despite going through lots of similar question related to this I still could not understand why some algorithm is susceptible to it while others are not.
Till now I found that SVM and K-means are susceptible to feature scaling while Linear Regression and Decision Tree are not.Can somebody please elaborate me why? in general or relating to this 4 algorithm.
As I am a beginner, please explain this in layman terms.
One reason I can think of off-hand is that SVM and K-means, at least with a basic configuration, uses an L2 distance metric. An L1 or L2 distance metric between two points will give different results if you double delta-x or delta-y, for example.
With Linear Regression, you fit a linear transform to best describe the data by effectively transforming the coordinate system before taking a measurement. Since the optimal model is the same no matter the coordinate system of the data, pretty much by definition, your result will be invariant to any linear transform including feature scaling.
With Decision Trees, you typically look for rules of the form x < N, where the only detail that matters is how many items pass or fail the given threshold test - you pass this into your entropy function. Because this rule format does not depend on dimension scale, since there is no continuous distance metric, we again have in-variance.
Somewhat different reasons for each, but I hope that helps.
Related
I'm working on a simple project in which I'm trying to describe the relationship between two positively correlated variables and determine if that relationship is changing over time, and if so, to what degree. I feel like this is something people probably do pretty often, but maybe I'm just not using the correct terminology because google isn't helping me very much.
I've plotted the variables on a scatter plot and know how to determine the correlation coefficient and plot a linear regression. I thought this may be a good first step because the linear regression tells me what I can expect y to be for a given x value. This means I can quantify how "far away" each data point is from the regression line (I think this is called the squared error?). Now I'd like to see what the error looks like for each data point over time. For example, if I have 100 data points and the most recent 20 are much farther away from where the regression line/function says it should be, maybe I could say that the relationship between the variables is showing signs of changing? Does that make any sense at all or am I way off base?
I have a suspicion that there is a much simpler way to do this and/or that I'm going about it in the wrong way. I'd appreciate any guidance you can offer!
I can suggest two strands of literature that study changing relationships over time. Typing these names into google should provide you with a large number of references so I'll stick to more concise descriptions.
(1) Structural break modelling. As the name suggest, this assumes that there has been a sudden change in parameters (e.g. a correlation coefficient). This is applicable if there has been a policy change, change in measurement device, etc. The estimation approach is indeed very close to the procedure you suggest. Namely, you would estimate the squared error (or some other measure of fit) on the full sample and the two sub-samples (before and after break). If the gains in fit are large when dividing the sample, then you would favour the model with the break and use different coefficients before and after the structural change.
(2) Time-varying coefficient models. This approach is more subtle as coefficients will now evolve more slowly over time. These changes can originate from the time evolution of some observed variables or they can be modeled through some unobserved latent process. In the latter case the estimation typically involves the use of state-space models (and thus the Kalman filter or some more advanced filtering techniques).
I hope this helps!
In sklearn, the document of QuantileTransformer says
This method transforms the features to follow a uniform or a normal distribution
the document of PowerTransformer says,
Apply a power transform featurewise to make data more Gaussian-like
It seems both of them can transform features to a gaussian/normal distribution. What are the differences in terms of this aspect and when to use which ?
It is confusing terminology that they use because Gaussian and normal distribution are actually the SAME.
QuantileTransformer and PowerTransformer are both non-linear.
To answer your question about what exactly is the difference it is this according to https://scikit-learn.org:
"QuantileTransformer provides non-linear transformations in which distances between marginal outliers and inliers are shrunk. PowerTransformer provides non-linear transformations in which data is mapped to a normal distribution to stabilize variance and minimize skewness. "
Source and more info here: https://scikit-learn.org/stable/auto_examples/preprocessing/plot_all_scaling.html#:~:text=QuantileTransformer%20provides%20non%2Dlinear%20transformations,stabilize%20variance%20and%20minimize%20skewness.
The main difference is PowerTransformer() being parametric and QuantileTransformer() being non-parametric. Box-Cox or Yeo-Johnson will make your data look more 'normal' (i.e. less skewed and more centered) but it's often still far from the perfect gaussian. QuantileTransformer(output_distribution='normal') results usually look much closer to gaussian, at the cost of distorting linear relationships somewhat more. I believe there's no rule of thumb to decide which one would work better in a certain case, but it's worth noting you can select an optimal scaler in a pipeline when doing e.g. GridSearchCV().
I am trying to generate a model that uses several physico-chemical properties of a molecule (incl. number of atoms, number of rings, volume, etc.) to predict a numeric value Y. I would like to use PLS Regression, and I understand that standardization is very important here. I am programming in Python, using scikit-learn. The type and range for the features varies. Some are int64 while other are float. Some features generally have small (positive or negative) values, while other have very large value. I have tried using various scalers (e.g. standard scaler, normalize, minmax scaler, etc.). Yet, the R2/Q2 are still low. I have a few questions:
Is it possible that by scaling, some of the very important features lose their significance, and thus contribute less to explaining the variance of the response variable?
If yes, if I identify some important features (by expert knowledge), is it OK to scale other features but those? Or scale the important features only?
Some of the features, although not always correlated, have values that are in a similar range (e.g. 100-400), compared to others (e.g. -1 to 10). Is it possible to scale only a specific group of features that are within the same range?
The whole idea of scaling is to make models more robust to analysis on features space. For example, if you have 2 features as 5 Kg and 5000 gm, we know both are same, but for some algorithm, which are sensitive to metric space such as KNN, PCA etc, they will be more weighted towards second features, so scaling must be done for these algos.
Now coming to your question,
Scaling doesn't effect the significance of features. As i explained above, it helps in better analysis of data.
No, you should not do, reason explained above.
If you want to include domain knowledge in your model, you can use it as prior information. In short, for linear model, this is same as regularization. It has very good features. if you think, you have many useless-features, you can use L1 regularization, which creates sparse effect on features space, which is nothing but assign 0 weight to useless features. Here is the link for more-info.
One more point, some method such as tree based model doesn't need scaling, In last, it mostly depend on the model, you choose.
Lose significance? Yes. Contribute less? No.
No, it's not OK. It's either all or nothing.
No. The idea of scaling is not to decrease / increase significance / effect of a variable. It's to transform all variables to a common scale that can be interpreted.
I have n points in R^3 that I want to cover with k ellipsoids or cylinders (I don't really care; whichever is easier). I want to approximately minimize the union of the volumes. Let's say n is tens of thousands and k is a handful. Development time (i.e. simplicity) is more important than runtime.
Obviously I can run k-means and use perfect balls for my ellipsoids. Or I can run k-means, then use minimum enclosing ellipsoids per cluster rather than covering with balls, though in the worst case that's no better. I've seen talk of handling anisotropy with k-means but the links I saw seemed to think I had a tensor in hand; I don't, I just know the data will be a union of ellipsoids. Any suggestions?
[Edit: There's a couple votes for fitting a mixture of multivariate Gaussians, which seems like a viable thing to try. Firing up an EM code to do that won't minimize the volume of the union, but of course k-means doesn't minimize volume either.]
So you likely know k-means is NP-hard, and this problem is even more general (harder). Because you want to do ellipsoids it might make a lot of sense to fit a mixture of k multivariate gaussian distributions. You would probably want to try and find a maximum likelihood solution, which is a non-convex optimization, but at least it's easy to formulate and there is likely code available.
Other than that you're likely to have to write your own heuristic search algorithm from scratch, this is just a huge undertaking.
I did something similar with multi-variate gaussians using this method. The authors use kurtosis as the split measure, and I found it to be a satisfactory method for my application, clustering points obtained from a laser range finder (i.e. computer vision).
If the ellipsoids can overlap a lot,
then methods like k-means that try to assign points to single clusters
won't work very well.
Part of each ellipsoid has to fit the surface of your object,
but the rest may be inside it, don't-cares.
That is, covering algorithms
seem to me quite different from clustering / splitting algorithms;
unions are not splits.
Gaussian mixtures with lots of overlaps ?
No idea, but see the picture and code on Numerical Recipes p. 845.
Coverings are hard even in 2d, see
find-near-minimal-covering-set-of-discs-on-a-2-d-plane.
A friend is using Factor Graphs to do text mining (identifying references to people in text), and it got me interested in this tool, but I'm having a hard time finding an intuitive explanation of what Factor Graphs are and how to use them.
Can anyone provide an explanation of Factor Graphs that isn't math heavy, and which focusses on practical applications rather than abstract theory?
They are used extensively for breaking down a problem into pieces. One very interesting application of factor graphs (and message passing on them) is the XBox Live TrueSkill algorithm. I wrote extensively about it on my blog where I tried to go for an introductory explanation rather than an overly academic one.
A factor graph is the graphical representation of the dependencies between variables and factors (parts of a formula) that are present in a particular kind of formula.
Suppose you have a function f(x_1,x_2,...,x_n) and you want to compute the marginalization of this function for some argument x_i, thus summing over all assignments to the remaining formula. Further f can be broken into factors, e.g.
f(x_1,x_2,...,x_n)=f_1(x_1,x_2)f_2(x_5,x_8,x_9)...f_k(x_1,x_10,x_11)
Then in order to compute the marginalization of f for some of the variables you can use a special algorithm called sum product (or message passing), that breaks the problem into smaller computations. For this algortithm, it is very important which variables appear as arguments to which factor. This information is captured by the factor graph.
A factor graph is a bipartite graph with both factor nodes and variable nodes. And there is an edge between a factor and a variable node if the variable appears as an argument of the factor. In our example there would be an edge between the factor f_2 and the variable x_5 but not between f_2 and x_1.
There is a great article: Factor graphs and the sum-product algorithm.
Factor graph is math model, and can be explained only with math equations. In nutshell it is way to explain complex relations between interest variables in your model. Example: A is temperature, B is pressure, components C,D,E are depends on B,A in some way, and component K is depends on B,A. And you want to predict value K based on A and B. So you know only visible states. Basic ML libraries don't allow to model such structure. Neural network do it better. And Factor Graph is exactly solve that problem.
Factor graph is an example of deep learning. When it is impossible to present model with features and output, Factor models allow to build hidden states, layers and complex structure of variables to fit real world behavior. Examples are Machine translation alignment, fingerprint recognition, co-reference etc.