Questions tagged [feature-construction]

Variables (used for prediction or explication) used in regression or regression-like models (like clustering, discrimination). Use this tag for questions about constructing such variables.

Filter by
Sorted by
Tagged with
1
vote
0answers
31 views

How to handle a feature set with unordered members?

I ask a group of people for the age of their 10 best friends, and will try to predict some output variable on basis of that. I don't ask them to rank these friends in any way, thus each person I ask ...
2
votes
1answer
35 views

Is it always possible to find the feature map from a given kernel?

Each positive definite kernel $k(x, x')$ used in machine learning/statistics has an equivalent representation as a dot product of the feature map representation $\phi(x)$ of each input i.e. \begin{...
0
votes
0answers
13 views

does the resolution of the vectors influence the result?

I have setup a model for prediction of free parking places. Thi model based on SVR (support vector regression) and sklearn libs. My feature vectors are: time as float (e.g. 8.00, 8.25, 8.5, 8.75, 9.0....
0
votes
0answers
8 views

How to turn a list of tags into features?

So, a while back I asked a question here and realised that I have to be smarter about my feature selection. To summarise, I downloaded some thousand pictures off the internet and graded them on a ...
2
votes
1answer
17 views

Ranking with multiple weights/ features

We have entities where every entity has start ($s_i$) and end ($e_i$) times and count $c_i$. An entity is important if its interval ($e_i - s_i$) is large and if its $c_i$ is large. Here's what I ...
0
votes
0answers
11 views

Influencing linear SVM features with artificial feature vectors to prevent overfitting (pseudo manual feature selection)

I'm training a linear SVM with about 20 features with the usual setup (10 fold cd, grid search, data standardized to mean=0, sd=1, positive class label=1, negative class=-1, balanced examples for both ...
1
vote
1answer
19 views

Kernel Confusion

Consider the function $K(\vec{x},\vec{y})$ where $\vec{x},\vec{y} \in \mathbb{R}^n$. I have been asked to check that this is a valid kernel. Question 1 My understanding is that I can prove this in ...
0
votes
0answers
14 views

PCA reconstruction for low rank high dimensional matrices

For an n x p matrix A, when n >> p with p ...
0
votes
1answer
21 views

Reconstruction error drops for an anomaly?

I have a convolutional Autoencoder being used as an anomaly detector, it works well. Today however I trained it on a new training/test data set and the anomalies were exposed as a drop in ...
1
vote
0answers
17 views

How does the “age” feature work in video recommendation systems?

In the paper Deep Neural Networks for YouTube Recommendations, it mentions that the “example age” feature helps recommending fresh contents in Section 3.3. Many hours worth of videos are uploaded ...
4
votes
3answers
155 views

Learning a target feature from data

I have a dataset of customers (infos about them, as well as their buying behavior) to whom ads are sent regularly. How can I design a target feature that will result in a good model that predicts when ...
0
votes
0answers
30 views

How to calculate High Order Moments?

I have a raw data set for Radio Signals (type: complex) and I want to calculate the High Order Moments and Cumulants (see this paper PART: II-A) In the Paper a ...
0
votes
0answers
15 views

Why can't I feature scale after encoding categorical data

I've read somewhere that feature scaling categorical encodings (with vector mean/variance or median/IQR) is a bad idea and breaks the structure of the encoding - something about orthogonality of ...
0
votes
0answers
23 views

How to combine multiple (facial recognition) feature vectors into one feature vector

I am writing a facial recognition library in C++ which performs inference to generate feature vectors. When I provide my machine learning model with image data, I receive a features vector of length ...
1
vote
0answers
19 views

For a regression model, can you transform all your features to linear to make a better prediction?

I was thinking. Would it be a good approach to check your features one by one (assuming you have a manageable amount of them) and see the relationship they have with your target variable, if they have ...
0
votes
1answer
18 views

feature engineering - dealing with conditional numeric variable

Let us say I have a dataset with 2 columns X1 and X2. X1 is dichotomous and has 2 level yes and no. X2 is numeric and is only completed when X1 = yes. Is it possible to keep the numeric column X2 as ...
1
vote
0answers
21 views

Is consistency among classifiers a good measure for feature sets?

I have two feature extraction approaches (or feature sets) to describe the same data, the feature set 1 has overall consistent results among 4 different classifiers (SVM, Logistic Regression, Adaboost ...
0
votes
0answers
10 views

Time units as features vs individual records

I am working on a project similar to predict likelihood of a student completing an online course given the performance to a given date. The approach taken so far is very similar to this https://...
0
votes
0answers
11 views

Feature engineering on hierarchically structured data

I have a regression problem to predict $y$ and a predictor categorical variable $x$ that has about $500$ different categories. $x$ is an occupation code of a company and is hierarchically structured, ...
1
vote
2answers
322 views

How to calculate Helmert Coding

I am trying to understand how Helmert Coding works I know it compares levels of a variable with the mean of the subsequent levels of the variable, but what are these levels and how can I calculate ...
6
votes
1answer
206 views

How would someone use curves as an input to a supervised learning model?

I was asked this question during a test and couldn't figure out the answer: You have a set of curves against time $X_i(t)$ that you want to use as input to a supervised learning model. The curves ...
0
votes
0answers
40 views

LSTM Autoencoders - architecture

I am a bit confused about the structure of LSTM autoencoders, as far as I know, common way to construct vanilla autoencoders is bottleneck structure, for instance, start with 40 nodes, encode it to 30 ...
0
votes
1answer
174 views

Multi-task learning: weight selection for combining loss functions

I am training a system that combines two sub-systems: one for classification and another for reconstruction. Can anyone suggestion what are the common practice for weight selection for combining two ...
3
votes
1answer
91 views

Is it valid to have all zeroes in a One-Hot Encoded categorical feature?

I'm building an MLP classification model and one of my features is the name of certain products. These names can be anything and in theory there could be an infinite number of different names in the ...
1
vote
2answers
94 views

What is the difference between overfitting and “not learning”

I am trying to build a Random Forests (RF) model using around 2000 observations and a number of features (can be 50 or can 1000, I still do not know which features are to be used). One way to ...
0
votes
0answers
23 views

How to properly add spatial features for a precipitation time series forecasting?

I am reading this paper. The center of the circle is the site where the model should forecast precipitation. Red stars in the picture are nearby sites and each site has these features: I want to ...
5
votes
0answers
37 views

Feature engineering for sheet music

I have a large dataset of digitized music scores that I'd like to use as input to a network. Initially, I'm looking to train networks to identify key signatures, tempo, dynamics, etc. from the raw ...
0
votes
1answer
47 views

Are legacy values useful for regression models?

I'm building a model that predicts house prices in order to learn some regression techniques. Currently I'm trying to engineer features that might be significant when predicting prices. I got a hold ...
0
votes
1answer
43 views

On starting feature engineering

I would like to start my feature engineering process by first selecting a subset of features that are highly correlated with the target feature. However, if I do select let’s say the top k in terms ...
0
votes
0answers
11 views

Finding Semantically Similar Learned Features

I have learned features from text and image and are projected in a hyperspace. Once I have the feature space, I am looking to find those features which are similar to each other. I have tried ...
0
votes
1answer
16 views

preparing free text column for regression

I have a column X which contains occupation/profession as an independent variable as free text, which is very much correlated with a continuous dependent variable. What techniques do you usually use ...
1
vote
1answer
265 views

SVD PCA reconstruction of data [duplicate]

I have some data about the $\{noise,~ size,~ speed,~ length,~ width\}$ of cars. I have performed SVD, and I want to reconstruct my data using only the first 2 principal components. I subtracted mean ...
0
votes
1answer
394 views

(Feature Selection) Meaning of “importance type” in get_score() function of XGBoost

I'm trying to use a build in function in XGBoost to print the importance of features. My code is like ...
2
votes
1answer
70 views

Interpolating principal component

In my thesis, I use PCA from a bunch of WVS responses to measure the social capital of a country (aggregating principal components to country averages). However, WVS provides a quite low frequency of ...
0
votes
0answers
53 views

Number of lags in Ljung-Box test for feature extraction

I'd like to cluster time series based on static features, one of which is the Ljung-Box autocorrelation. After reading this question on "How many lags to use in the Ljung-Box test of a time series", I'...
2
votes
1answer
67 views

What's the term used when identical feature vectors map to different target variables?

Context: Fitting a Machine Learning Algorithm on a labeled dataset. For a feature vector [a,b,c] and a labeled output/target variable, what's the term used when identical feature vectors map to two (...
0
votes
0answers
31 views

Classifying matching object pairs with Python / sklearn

I have a ground truth dataset of around 600,000 objects. Each object has 100 features, and for each pairwise combination of objects, I have a 0-1 relationship ("equal" / "not equal"). What is the best ...
0
votes
1answer
39 views

Using outlier records as a feature in model building

I am exploring the Big Mart Sales III dataset and trying to understand if using outlier rows to build a feature for predictive modeling is a sound and correct approach. This is how I have proceeded ...
1
vote
0answers
24 views

Learning useful semantic representations of data

Training a neural network on its final task (e.g. classification) right from the beginning is not always the best way to go. I'd like to make a short list of recognized methods of motivating a NN to ...
0
votes
0answers
15 views

Data Representation for sequential input NN

The essence of the problem I want to model is to go from input sequences(length n) to a "distance matrix"(n x n). Although for this distance matrix we are not actually predicting $n^2$ values because ...
1
vote
1answer
108 views

What is the definition of a “rectified conv feature map” in a convolutional neural network mentioned in the paper of “visual explaination”?

I have read the answer of the question What is the definition of a “feature map” (aka “activation map”) in a convolutional neural network? But I don't think that it is same as what I want. I ...
1
vote
1answer
121 views

When using linear function approximation how (and why) should I incorporate the actions into the feature vector?

When reading R. Sutton: Reinforcement Learning - An Introduction (2nd edition), in chapter 10.1 Episodic Semi-gradient Control, the Mountain Car problem is mentioned and as an example it is solved ...
0
votes
1answer
139 views

How to engineer a bimodal continuous feature for use in Decision Tree?

I have a predictor that exhibits "bimodal" behaviour. How can I engineer this feature to improve performance within a Decision Tree? For an intuitive example, consider how a binary flag of "moves ...
0
votes
0answers
10 views

Predicting based on regressor measured over time

Suppose I want to predict whether a patient has post-operative complications. In addition to some 'usual' regressors, such as age and weights, I also have access to variables that are measured over ...
0
votes
0answers
29 views

How to extract static program features automatically?

I did want to know how to extract statistical features from program. Like supposing I wanna do an extractor for loops programs so features in this case could be The loop nest level. Is the loop ...
0
votes
0answers
42 views

Sparse coding and feature learning

Recently I tried to understand sparse coding and its application to classification. But there is no way to check whether I understand correctly, so I have a few questions about this algorithm. I ...
0
votes
1answer
38 views

Algorithms for Graphs Clustering

Which methods is available for graph clustering? The most information by query "Graph clustering" concentrated on the finding set of nodes in the one large graph (or graph partition), but it isn't my ...
2
votes
1answer
307 views

Feature Engineering: Should I drop features that can be calculated using other features?

In feature engineering, should I drop all features that can be calculated using other features? For example, let us say that we have this dataset: ...
0
votes
1answer
60 views

Addressing “Unbalanced Features” or Feature Taxonomy for Nearest Neighbor / Similarity Calculations

The main question is how to address an imbalance in representation of feature "sets" when calculating similarity. I'll motivate with an example scenario: Suppose we have objects described by a binary ...
0
votes
0answers
49 views

State of the art in feature extraction from review text

I am working on a sentiment review classification problem and so far i have explored POS tags, synsets, N-grams, word2vec, tf-idf, doc2vec, glove and fastext vectors as features. I am wondering what ...