Questions tagged [feature-engineering]

Feature engineering is the process of using domain knowledge of the data to create features for machine learning models. This tag is meant for both theoretical and practical questions regarding feature engineering, excluding questions asking for code, that would be off-topic on CrossValidated.

Filter by
Sorted by
Tagged with
1
vote
1answer
15 views

Encoding cyclical feature minutes and hours

I'm working with time-series data to train a binary classification model that predicts if an event is going to happen or not in the future. The likelihood of the event depends on the specific time ...
0
votes
0answers
10 views

Time-dependent feature analysis

I have a linear relation between variable A and variable B. Variable A is an area under the curve, where the curve is a gaussian fitted to a time series evolution. Now, apart from the time series data ...
0
votes
0answers
8 views

How to use target encoding : expanding mean on the test set

The expanding mean is a way to prevent overfitting when performing target encoding. But what I do not understand is how to use ...
0
votes
0answers
19 views

modeling a electrical pulse which is technically time dependent

I have an electrical pulse that I need to fit a curve to a certain area of but not the entire thing. The whole pulse looks like this However the only part that I need to model is this My boss ...
1
vote
0answers
20 views

Feature engineering: including counter-parties of a transaction in a dataset

Background Say I have a dataset of transfers between bank accounts structured like so: ...
2
votes
2answers
221 views

Difference between Feature engineering and hyperparameter optimizations?

Hyperparameter optimizations and feature engineering can(in my understanding) both be used to create a machine learning model. But what is the difference? And what is done to the y = wx + b formula in ...
0
votes
0answers
10 views

Extracting Features to Determine Periodicity

I have accelerometer time series data sampled at 30Hz from participants and am extracting features from each separate movement in the collection period per person to use for machine learning. I have ~...
0
votes
1answer
20 views

How to handle potential ambiguity when one-hot encoding?

Let's say I have two categorical features: Movie, Director. I one-hot encode both the Movie and Director features for use in a linear regression model. The problem is that two or more movies may be ...
1
vote
1answer
19 views

Does mislabeling due to adversarial noise in features count as adversarial machine learning?

According to the traditional definition, Adversarial machine learning is a technique employed in the field of machine learning which attempts to fool models through malicious input. However, I have ...
1
vote
1answer
21 views

Different scales of input features for stacking ensembles?

I have two models to predict future stock market behavior based on historical data: ARIMA time series model lstm model (including data from various other sources) ARIMA tries to model the daily ...
1
vote
0answers
20 views

Can RFs find a product interaction between two independent variables?

I'm doing the FastAI course on ML, and the main topic that is currently being discussed is random forests. Jeremy Howard explains how random forests, unlike something such as logistic regression, can ...
0
votes
0answers
18 views

Combine TFIDF with non-textual features

I am dealing with an email classification problem in which I have email requests coming from different groups of people. I am building a classifier to classify these emails based on historical email ...
1
vote
0answers
32 views

Machine learning: use benchmark as a feature

The project I am doing is to predict surgery lengths. The benchmark I am trying to compare is to take the average of most recent 20 cases for the cases with the same ID. What I tested is to use this ...
0
votes
0answers
13 views

The most basic question about Feature Important or Permutation_Importance

Consider the XOR gate with three inputs. The truth table will be: Now all the variables on their own are near random as far as the model is concerned. Each input 1 or 0 has a 50% chance of being ...
0
votes
0answers
16 views

Reduction in number of observation by extracting piecemeal signal features,while keeping the no of features same. Can it be called feature extraction?

I have a dataset generated from 9 sensors in an E-nose system for a binary class classification problem. The system provides a response for 240 seconds for each sample. i.e. I have a data set of 240 * ...
1
vote
1answer
64 views

“Deep learning removes the need for feature engineering”?

I have seen it written in several papers and currently see it written in Deep Learning with Python by Francois Chollet that Deep learning removes the need for feature engineering What does this ...
0
votes
0answers
19 views

Should I engine features from coordinates (positional data)?

I am trying to do a regression on housing price (price/m^2). Apart from the lat and lng of the property, I also have city_code, district_id, street_id. I am thinking whether I should remove city_code,...
1
vote
0answers
11 views

Imputing null values for metrics used as features in ML model

I have a data set from a live application. Each row is a user interacting with the app. We are predicting a feature for which we currently have a deterministic solution for. There is ample training ...
0
votes
1answer
26 views

How to deal with over-represented features in anomaly detection

Let me illustrate the problem with an example: Our dataset consists of a collection of letters written to Santa Claus mostly by kids together with the age of the person who wrote them. We want to run ...
0
votes
0answers
41 views

How do I create a classification model to predict observations with different feature vector lengths?

Say I have a dataset containing hourly records of the vital signs of people trying to survive in the wild, their environmental conditions, and a label of whether the person survives. I would like to ...
1
vote
0answers
7 views

Measuring the effect of adding a feature descriptor?

I have a class A that is often classified as class B. I use Local Binary Patterns in conjunction with a SVM. If I concatenate a color histogram to a standard Local Binary Pattern, it increases it's ...
1
vote
0answers
18 views

Is it good idea to generate features from data points similarity comparison?

I know about polynomial features in machine learning, which can introduce nonlinearity to original dataset. I also heard about binning, which also allows us to create new features from existing ones. ...
1
vote
3answers
52 views

ML technique recommendations where each feature has multiple properties and the number of features per observation varies

Here is an abstracted version of the problem I am facing. I wish to predict the true value of a variable from multiple noisy predictions as to its value. There are three complicating factors: For ...
0
votes
1answer
27 views

is it a good idea to take the derivative or integral of some features and add them as new features in machine learning?

I'm learning how to do feature Engineering and come across some ideas in my head that's why I want to ask if I had some dataset with some features let's say 2 features and I have a timestamp column ...
0
votes
0answers
16 views

Model Examples always with 1 or 2 features

Why are all the model examples that I see on sklearn (e.g., https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.LocalOutlierFactor.html or https://scikit-learn.org/stable/auto_examples/...
0
votes
0answers
14 views

I have capped my response variable, should I calculate my RMSE/MAE/MAPE with the true values capped or not?

So, I have trained a model in my train set with the response variable with a superior limit. Because, the peaks are not important for my analysis. And if I dropped it, I would lost a lot of data. ...
0
votes
0answers
5 views

Using supply as feature in price predictor model

On a machine learning model that outputs the optimum price of a product (ex: cars listed on some website), would it make sense to use the number of instances of that product as a feature? In the case ...
1
vote
1answer
33 views

Combining continuous and binary data in unsupervised learning

I am working on cluster detection in a data set consisting of housing data. Each data point has some continuous features, such as house size, and some discrete ones, such as the number of garages (0 ...
0
votes
0answers
21 views

Scaled and Unscaled features are giving different feature importances when using LIME

Now, based on my understanding, feature scaling should have no impact on my model results due to the fact that XGBoost isn't sensitive to monotonic transformations. Ref My concern is the model ...
1
vote
0answers
36 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 ...
0
votes
0answers
24 views

Validity of PU learning when using character-level encoding and CNNs for text data

I'm trying to classify a large set of documents (~100M) as valid or invalid, based upon a small given set of labeled valid documents (~3k). I'd like to know if the PU learning approach described in ...
0
votes
1answer
26 views

What are the steps and correct order of the operations in Machine Learning? [from Getting data to optimising models]

I've followed lots of tutorials on Machine Learning but in each of these, they go for a different strategy so it's quite confusing for me. I want to Know that what are the operations involved and what ...
0
votes
0answers
11 views

Effect of feature interaction in classification model

I have a binary classification problem. I have checked below things as EDA- 1) Distribution of features to see any outlier, linear model can be fit? 2) Box plot to see which variables will be ...
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 ...
0
votes
0answers
21 views

How to identify important features in deep learning models

I am mostly familiar with traditional hand-crafted feature setting where we use ML algorithm such as SVM to analyse these features. In this way, we can identify what were the most important features ...
0
votes
0answers
11 views

Handling percent change variables with 0s

I'm building a model, and adding in some percent change features. I'm running into an issue when calculating percent change month over month. Let's say I have a product and I'm just calculating the ...
2
votes
1answer
18 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
22 views

Feature with very few extreme values

As a set of inputs into a machine learning model with ~10 million samples, I have a feature that exhibits a pretty extreme shape. Most of the values (98.5%) are 0, but those that aren't 0 tend to be ...
1
vote
0answers
25 views

Missing Data imputation on one continuos column which depends on another feature and which does make sense only when such feature is positive

For each row (open contract) of my dataset, I have got a certain number of orders. I have created some features related to such orders; let's take for instance the average and the std deviation of the ...
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 ...
0
votes
0answers
8 views

How to model dependency between input features when building a classifier

I have dataset with the shape of (1000,20) (1000 rows, 20 features) and I want to build a classifier for it. However, most sk-learn algorithms assume the these 20 features are independent. In my ...
0
votes
0answers
13 views

Including ratios as features in machine learning algorithms [duplicate]

Assume that body mass index (BMI) is a good predictor for early death which we are trying to model using a host of different algorithms. A colleague posed the following to me: would it be better to ...
0
votes
1answer
29 views

Add Absolute Value as a Feature

In a machine learning context, does it make sense to have both a measurement and its absolute value transformation as features? There are already ~120 features in this predictive model (an elastic ...
1
vote
1answer
46 views

Basic question about feature selection

I am new to machine learning. I have a basic question about feature selection. I have a dataset with 100 features which I used to regress an output Variable. When I do regression with all the ...
1
vote
0answers
26 views

How to treat low frequency continuous variable in machine leanring

Hello I am working on machine learning model for count data, and I have various features that are highly skewed. The frequency table for one of the feature is given below. ...
0
votes
1answer
23 views

Training Dataset vs Feature

What is the difference between training data and features in a machine learning model? Is feature just building blocks of training dataset?
0
votes
0answers
9 views

how to extract features when observations and feature set are almost equal?

I need to do a regression where I have in total 750 observations and nearly 500 features. I am struggling to come up with feature engineering technique that gives top 10 or 15 non col-linear features ...
0
votes
0answers
74 views

Encoding Ordinal categorical data using Python

I am trying to encode ordinal data. I found a post which suggests a way to do it. Where to find a guide to encoding categorical features? This seems to make sense. For nominal data, I would do the one-...
2
votes
0answers
21 views

Variable selection, variable reduction, and handling sparsity for binary text classification

I am trying to do a binary text classification using support vector machine. I am wondering if I am doing it right and I'd like to look for some answers to the questions in mind. The following ...
0
votes
1answer
62 views

Decide for three Bernoulli samples are some of them from the same distribution (or non of them) ? (do not use t-stat ) [duplicate]

Please pay attention, I am interested in Bernoulli samples, and hope to find criteria specific to Bernoulli distribution, not using s Student's t-statistics or Mann-Whitney or etc., since their use ...