Methods and principles of selecting a subset of attributes for use in further modelling

learn more… | top users | synonyms (2)

1
vote
1answer
71 views

Scale-invariant feature transform explanation

How do I explain the scale-invariant feature transform (SIFT) to a layman?
0
votes
0answers
20 views

I observed very different feature scoring from two different classifiers. What does it really mean?

Here what I've done. Given the dataset, I run a Random Forests and Logistic Regression with 5 Fold Stratified Data Sampling. Then I plot the feature importance for Random Forests and Logistic ...
2
votes
1answer
39 views

Automatically fixing ill-conditioning or collinearity

I'm backtesting a regression model, which entails running it on a bunch of bootstrap samples of a "rewound" version of our data set. Unfortunately, in some of these resamplings, I end up getting some ...
0
votes
0answers
26 views

Appending additional data to learnt autoencoder features

My task is to perform image segmentation / full scene image parsing. I am working on an outdoor dataset which was taken under strict spatial constraints. The images contain fruit on trees and the ...
1
vote
1answer
110 views

No significant tests when using Benjamini-Yekutieli multiple testing correction on millions of tests

I am using a univariate filter to reduce the number of features prior to applying a learning algorithm to a huge binary classification dataset (22510066 features x 500 examples). All the features are ...
2
votes
0answers
92 views

Deep learning: representation learning or classification?

For classification, I have often heard about deep learning / deep neural networks as a form of representation learning. I am confused as to what "representation learning" means in this context. Which ...
2
votes
0answers
47 views

Reference for this claim: important features in data can be “hidden” in the higher PCA axes that are typically thrown out [duplicate]

I remember reading a paper a while ago that demonstrated some cases in which PCA would fail to capture important features of a data set in the first few principal components, but where those features ...
0
votes
2answers
25 views

R feature selection [duplicate]

I am working with the method randomForest for model building. And for a good model performance, it is very important to select the right features. At my example I have 30 variables and I would like to ...
0
votes
0answers
12 views

How can be assesed that a given data representation is better than the other?

Given a classification dataset, suppose I learn many different data representation with Matrix Factorization, Clustering or with such approaches. At the end , how would I decide which is better than ...
0
votes
0answers
31 views

Data-driven, high-dimensional feature selection strategies

I am working on a biomedical/healthcare data science problem. I have a dataset of 600 samples, ~6000 variables and class label as "positive" or "negative". I want to perform feature selection on ...
0
votes
1answer
52 views

Classifier with variable number of features

I am trying to make a classifier when each sample has a variable number of features. An example of how this could occur is, for example, if the features are the purchases (type, dollar amount, etc) ...
1
vote
2answers
106 views

How To Better Represent A Problem To A Machine Learning Algorithm

I am familiar with the basics of how to present a problem to a machine learning algorithm using binary encodings. I am also familiar with, but still learning about, feature selection/extraction and ...
0
votes
0answers
25 views

Find the relative importance of features weights in multi-class SVM without PCA - plotting coef distribution?

I'm classifing users with a multiclass svm (one-against-on), 3 classes. In binary, I would be able to plot the distribution of the weight of each feature in the hyperplan equation for different ...
0
votes
0answers
31 views

feature slection in random forest in python

I have a dataset consisting of 24 numeric features and about 7000 rows, i am applying random forest to get the binary classification, So please tell me how to find only the relevant features to get ...
0
votes
0answers
12 views

What's the probability that there exists a hyperplane that can split a dataset which have random feature values ?

Given n data points, each with d features, n/2 are labeled as -1, the other n/2 are labeled as 1. Each feature takes a value from [0,1] randomly (uniform distribution). What's the probability that ...
1
vote
1answer
106 views

Justification for feature selection by removing predictors with near zero variance

I have a large number of variables that I'm trying to reduce, and I've stumbled on Kuhn's (2008) suggestion that I eliminate variables with zero or near-zero variance. This makes sense to me, it's ...
2
votes
1answer
54 views

Will adding additional features hurt the performance of SVM ?

Just wondering the effects of additional features. Following are several thoughts: If the additional features are noisy (can not distinguish the two classes), then additional features won't hurt ...
1
vote
1answer
27 views

What kind of general strategy can you apply after selecting model and hyper parameter training?

As a rookie to machine learning area, I tried to play some Data Science tutorials and beginner competitions to gain some knowledge and experience. The problem I encountered in every scenarios is ...
1
vote
0answers
16 views

Relation between chi-squared statistic scores and classification accuracy

I am evaluating the utility of two distinct sets of features for solving a given supervised classification problem with two classes. I am using the chi-squared statistic as a feature selection ...
0
votes
0answers
17 views

Finding the features that have meaning in subset of data

I have a set of $N$ points $x_i=(x_i^1, x_i^2,...,x_i^{m+k})$ in $m+k$-dimensional space ($m$ continuous dimensions and $k$ discrete). Also I have a subset of these points that are marked as "bad". ...
0
votes
1answer
58 views

After adding additional features, same accuracy on test data, but higher accuracy on training data, how should I interpret ?

I've done 5-fold cross-validation and the model is SVM. 300 features: 0.53 on test, 0.55 on training; 700 featuers: 0.53 on test, 0.67 on training. Does this mean that the additional 400 features ...
2
votes
1answer
208 views

Feature selection for time series data

I am looking for methods for feature selection (or feature extraction) for time series data. Of course I did some research before, but it was not satisfying. I am aware of methods like PCA, ...
1
vote
0answers
32 views

High dimensional explanatory variable

I have a data set of 22 observations and 6931 variables. Data belongs to two classes, 0 and 1. I would like to know which features are important for each class (species) and which one contribute the ...
1
vote
0answers
28 views

Learning if instances from a dataset are part of the same subset

I was wondering if there are some well-known machine learning methodologies for subset learning. In other words, to learn if two instances are part of the same subset or not (boolean label?). One ...
2
votes
1answer
197 views

How to increase the performance of random forest classifier?

I have a text classification task. These are the metrics for different languages at present: class1: 0.6823 class2: 0.7450 class3: 0.66 class4: 0.6719 How can I ...
0
votes
1answer
304 views

R gbm package variable influence

I'm using the excellent gbm package in R to do multinomial classification, and my question is about feature selection. After deciding the number of iterations ...
0
votes
0answers
20 views

Feature Selection for Regression Models in R [duplicate]

I’m trying to find a feature selection package in R that can be used for regression. Most of the packages implement their methods for classification using a factor ...
0
votes
1answer
165 views

Feature selection in GBM

I am using gradient boosting (caret package in R). As far as I understand, the feature selection is already included in this package. However, I slightly ...
8
votes
4answers
206 views

How can top $k$ principal components retain the predictive power on a dependent variable?

Suppose I am running a regression $Y \sim X$. Why by selecting top $k$ principle components of $X$, does the model retain its predictive power on $Y$? I understand that from ...
1
vote
1answer
160 views

GBM: Predict the response variable measured in {0,20}

I need to predict the response that has values in {0,20}. Should it be used as a factor or as a numeric value? How does it influence on the prediction error? I am using GBM with the Gaussian ...
1
vote
0answers
12 views

Need a statistic for comparing “strength” of Markov blankets in a Bayesian network

Working with Bayesian networks. I take a given network structure and fit its parameters on data. I am looking for a statistic based on those parameter estimates that allows me to compare Markov ...
0
votes
0answers
30 views

How many samples are enough

I have objects with large number of attributes (about 60.000). Attributes are actually deviations of object part from model. I would like to cluster this objects, to get centroids that will represent ...
0
votes
1answer
83 views

Find entropy in WEKA

I am new in data mining so sorry for asking this kind of silly question. I am working on FAST feature selection algorithm and for that I need to find entropy of each attribute in dataset. But the ...
1
vote
0answers
26 views

feature selection in a small sample size

I need an advice. I have a dataset consisting of 108 observations (27 subjects * 4 time points) and ~10000 features. The data represents intensity values (comes from continuous domain). When I run ...
6
votes
4answers
228 views

How to prepare/construct features for anomaly detection (network security data)

My goal is to analyse network logs (e.g., Apache, syslog, Active Directory security audit and so on) using clustering / anomaly detection for intrusion detection purposes. From the logs I have a lot ...
3
votes
2answers
141 views

Appropriately selecting explanatory (independent) variables

My aim is to carry out a GLM. I have 400 sites where I have count data of animals (response variable) and environmental characteristics (explanatory variables). At the moment I have around 40 ...
0
votes
0answers
16 views

What must be the sample size for feature selection by coefficient correlation method?

I have eight features from which I want to select 2-3 significant features for classification. The method which I adopted for doing so is coefficient correlation. The problem I am facing is for ...
1
vote
0answers
12 views

Will the classification accuracy vary if we first classify based on a single variable and then use the rest?

Let's suppose I am doing classification and that I have 99 features and another feature that says if the person is male or female. I have two options viz to build one classifier using all the ...
4
votes
1answer
124 views

What are the disadvantages of using Lasso for feature selection?

As far as I understand, feature selection is difficult for classification problems because it's effectively impossible to identify an optimal subset of $k$ features in problems where the the total ...
1
vote
2answers
221 views

Feature selection before neural network classification

I have a training set of 87 samples and 9480 variables. My predictors are continuous and my response variable is binary. I'd like to use the caret package in R to tune a neural network classification ...
0
votes
2answers
55 views

Which variables to keep in my analysis based on loadings from PCA? [duplicate]

Could someone please explain me how I should decide which variables to keep in my analysis based on loadings from PCA. The output is: ...
1
vote
0answers
45 views

How to reduce the number of features for Gaussian Process regression?

Ridge regression reduces complexity of the model by scaling down the coefficient. Lasso reduces the complexity of the model by selecting the features used. For Gaussian Process, is there similar way ...
1
vote
1answer
103 views

Quantitative importance for interacting variables in Artificial Neural Networks?

Is there any common/sound method to quantify (similar to T-test or F-test in regression models) the measures of influence and significance of terms in Artificial Neural Networks? By terms I mean both ...
1
vote
0answers
42 views

Akaike information criterion for categorical and numerical data

How should I compute AIC for categorical and for numeric variables in classification problems? I see in Chapter 6 of Zumel and Mount that they use AIC before they train classification algorithms ...
1
vote
1answer
40 views

Alternative to AIC for feature selection in classification

I want to know what are the most common methods for feature selection in classification problems (binary and mutli-class). I see in Chapter 6 of Zumel and Mount that they use AIC before they train ...
0
votes
0answers
21 views

How to find original features corresponding to the first two principal components? [duplicate]

I have a set of data described by $n$ features. I do a principal component analysis (PCA) to reduce it to just 2 dimensions so I can make a 2D plot of the data, with the first two Principal ...
3
votes
1answer
115 views

When should I use feature selection and when should I use dimensionality reduction techniques?

When should I use feature selection and dimensionality reduction? I know that feature selection is different from dimensionality reduction. But I don't know under what circumstances should I use ...
0
votes
1answer
158 views

Model Selection and RFE using caret

I'm faced with a high dimensional (samples=148, features=20000), supervised binary classification problem. Which I would like to approach with an ensemble of classifiers, that will classify using a ...
0
votes
0answers
42 views

In PCA, can the values in the principle component vectors which are close to zero be removed to see the important features? [duplicate]

In PCA, when I extract the principle component vectors, I am choosing the first vector with the largest corresponding eigenvalue. I notice that some of the values in this vector are close to zero. Can ...
0
votes
0answers
23 views

To be significant or to be stabile, what's more scientfically important?

Recently I discovered the techniques related to cross validation. Basically you can split up your data in n groups and then run your model on one part of the data and assess prediction reliability on ...