Data organized into discrete categories or *classes* may present problems for certain analyses if the number of observations ($n$) belonging to each class is not constant across classes. Classes with unequal $n$ are *unbalanced*.

learn more… | top users | synonyms

0
votes
0answers
7 views

Logistic GEE Models with Extreme Class Imbalance

I have two questions Question 1: I have a longitudinal data with extreme class imbalance: 9 cases of the event of interest out of a total of 3000. I believe I have 3 options and this is the of ...
2
votes
2answers
132 views

Why P>0.5 cutoff is not “optimal” for logistic regression?

PREFACE: I don't care about the merits of using a cutoff or not, or how one should choose a cutoff. My question is purely mathematical and due to curiosity. Logistic regression models the posterior ...
1
vote
1answer
16 views

What is the effect of training a model on an imbalanced dataset & using it on a balanced dataset?

When evaluating a model, for example a binary classifier, should the train and test set have 50% + and 50% - label distribution or could the distribution be random? If the distribution is biased in ...
0
votes
0answers
9 views

Kappa for classifier evaluation with ground truth data

for a fraud detection (credit-scoring-like) problem I want to use kappa as an evaluation metric. Kappa has the nice property that it fits very well to a problem with highly imbalanced relation between ...
2
votes
1answer
47 views

Unbalanced linear mixed effect modeling for longitudinal data with lme4

I'm new to longitudinal analyses, and I'm having trouble formulating a model that accurately reflects my study design. This study recruited subjects for two groups (dx vs. control), with measurements ...
0
votes
1answer
13 views

Suggestions on choosing correct method for multi-class classification of imbalanced data

I have data that is split into three classes (A, B and noise). The data amount is around 10000 samples, and A and B is only less than 5-10% of data. What is the best approach to handle this situation ...
3
votes
1answer
34 views

Predict probability when model was trained in balanced dataset

I have a dataset of about 1M observation and I had to predict a response that occurs only about 10.000 times (1%). I decided to train a random forest, but this takes a lot of time to train because ...
0
votes
0answers
12 views

Phenomena of undersampling the dataset

I have a question regarding undersampling the dataset. Under-sampling it's well known technique when you remove instances of the over-represented classes from the dataset until every class has equal ...
0
votes
1answer
30 views

Class imbalance in clustering

Is there is a problem for clustering if the dataset is highly imbalanced? I have a clustering task and it looks like that there is a realy huge peak whose tail covers other clusters. Are there any ...
1
vote
1answer
19 views

How to design a train and test set from a labeled dataset with class imbalance?

The labeled dataset I am using is almost 80% positive examples, 20% negative examples. However, I do not know the distribution of the data fed into the classifier. In this case, does it make sense ...
1
vote
1answer
19 views

Does class weighting introduce bias in Random Forest classifier?

I want to use a Random Forest classifier to stratify a strongly imbalanced population of samples. During training I used class weighting to weight the vote for each class by considering its ...
1
vote
1answer
46 views

Classification/evaluation metrics for highly imbalanced data

I deal with a fraud detection (credit-scoring-like) problem. As such there is a highly imbalanced relation between fraudulent and non-fraudulent observations. http://blog.revolutionanalytics.com/2016/...
0
votes
1answer
16 views

Where to start on understanding Adaboost for an unbalanced multi-class problem

I need a better classifier for an unbalanced multi-class problem. I'm tempted to implement Adaboost as an exercise in understanding it since it is quite clear and simple, but am unsure how to deal ...
0
votes
0answers
7 views

Setting effect weights to sum-to-zero in linear mixed effects model with unbalanced data

I am fitting a linear mixed effects model to longitudinal data. There is a between-Subjects factor Group with three levels (Info,...
0
votes
0answers
14 views

Learning from Multiple Naive Annotator's Presence (and their labels as Y)

Although I am new to the forum, I have been following for several years and wanted to begin by sincerely thanking everyone, from the administrators to the members, for creating such an amazing ...
0
votes
0answers
47 views

Recursive feature elimination and class imbalance

I am trying to apply the recursive feature elimination in the R package caret following the example in the caret website: ...
0
votes
0answers
33 views

What is the effect of class weight in Random Forest/Extra Trees Classifier variables importance?

In the sklearn implementation of random forest and extra trees classifier a class_weights parameter is available http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.ExtraTreesClassifier....
0
votes
0answers
22 views

How to analyze differences between two time series collected under two different conditions

How should one analyze time series data that were collected during two seasons: season 1 and then at season 2, where during each season a measurement sample for several continuous variables were ...
0
votes
0answers
26 views

Learning with unbalanced classes

I have a biometric data set of size $n \sim 150,000$, with $p=21$ features $x_1,\dots,x_p$ representing information about people $-$ some categorical and some numerical. The original data set had only ...
0
votes
0answers
13 views

Literature (guidelines) on unbalancedness in two-way within-subject ANOVA

I am looking for literature (guidelines) which discuss the consequences of unbalanced designs on running a (two-way) within-subject ANOVA and pros/cons of various counter-measures. I came up with ...
0
votes
0answers
7 views

Rebalancing doesn't give good results

We have a given a group of customers a price increase and as a result some have cancelled. We can identify exactly who they are because of the way they cancel. This means I can split them out from ...
1
vote
1answer
68 views

Linear regression with unbalanced dummy variables + not normally distributed residuals

I am conducting a multiple linear regression analysis in SPSS. My DV is a score between 0 and 6, and my predictors are: one dichotomous nominal variable (native vs. non-native speakers) one ...
0
votes
0answers
23 views

Unbalanced Repeated Measures Anova

First Question: I would like to double check that my code is doing what I think it is doing. I have two independent variables, Previous drilling and Soil Type, that are observational. Simpsons ...
1
vote
0answers
14 views

How to retrain a production classifier that blocks its own positive examples?

I'm looking for help understanding how to re-train a fraud detection classifier that's been deployed to production (where it successfully blocked much, but not all fraud coming into the system). I ...
0
votes
0answers
24 views

What to do with highly unequal sample sizes in a dummy linear regression? (+ nonnormal distribution)

I am conducting a multiple linear regression analysis in SPSS. My DV is continuous (score between 0 and 6), and my predictors are: one dichotomous nominal variable (native vs. non-native speakers) ...
0
votes
0answers
41 views

How to add hard negatives to original training data?

I have 2 class binary classification problem with original training data of size N=n_pos+n_neg in general case n_pos!=n_neg but now we can assume that number of positive and negative examples near the ...
2
votes
0answers
21 views

How to set sum contrasts for unbalanced factors

Let's say that I have a model where the response time depends on accuracy (0/1, coded either as categorical or numerical) and another categorical variable (pres: idem/diff), both interacting with the ...
0
votes
0answers
65 views

Caret classification: feature selection & unbalanced data

I have a two-class classification problem with very unbalanced data (~1:1000 Yes/No ratio). The initial model class I'd like to try is regular glm. So there are two issues need to be addressed: 1) ...
0
votes
0answers
24 views

What is the method to find difference in mean of test and control population where data were collected for a marketing campaign?

What is the method to find difference in the response rate of test and control populations in SAS where data were collected for a marketing campaign? (10-20% were control, and the rest were the test ...
0
votes
0answers
8 views

How to do posthoc comparisons for unbalanced 2-way ANOVA (type II SS)?

I am using the car package to perform a type II ANOVA on unbalanced data. My two factors are "storm size" and "storm frequency." I have two storm sizes and four storm frequencies. I only have both ...
0
votes
0answers
12 views

How to create stratified subsets of one file?

I have one large file with class imbalance problem. I would like to stratify the subset into 10 subsets, and to preserve the ration of class sizes for each fold. So for example the overall class ...
0
votes
1answer
21 views

Balancing classes for Neural Network training

In a speaker recognition problem I have 330 speakers (classes) as targets and want to predict the identities with a feedforward neural net with a softmax output layer. The thing is some classes have ...
0
votes
0answers
26 views

Using priors, weights or costs for mitigating class imbalance?

A plethora of Matlab classifiers (e.g. tree-based or svm) allow to set priors, costs or weights for the data points. This can help dealing with imbalanced data. Unfortunately, none does support ...
1
vote
0answers
24 views

Tuning priors/weights/costs to counteract class imbalance

I have a classification problem which consists of two classes. It has high class imbalance. There are around 85% data points for the negative class and only 15% for the positive class. One option is ...
0
votes
0answers
11 views

Group treatment with unbalanced repeated measurments

In this study I want to determine if treatment group b and/or c are different from control group a. There are 13 individuals in the study. The groups are unbalanced as there is a different number of ...
0
votes
0answers
22 views

Is this a 2 fixed unbalanced ANOVA? How can be tested normality and homoscedasticity?

I need to know if my biological experiments show discrepancies between the condition used. In my experiments I have 2 fixed conditions: type of substrate (2 types) and chemical added (1 control + 3 ...
0
votes
0answers
13 views

Class imbalance and standard errors

I'm building a logistic regression that models the probability of conversion when clicking on a website ad. I'm not that interested in building a great classifier, but I want to identify a set of the ...
0
votes
1answer
43 views

When is dataset considered unbalanced?

I have data set which is highly unbalanced - target attribute is 93% False and 7% True. But I know that this is normal for my kind of data. I am afraid that if I undertake any steps (I can take less ...
0
votes
1answer
30 views

Random Forests with modified partitioning criteria

Here is the context of my question : I'm doing binary classification with unbalanced classes. The measure of performance I'd like to maximise is a modified F-measure : $$ F_{\alpha} = \frac{1}{\frac{\...
3
votes
0answers
93 views

multiclass unbalanced data

I am trying to predict crimes (san francisco) using machine learning algorithms. Its a multi class classification problem with unbalanced data. I took sample of data ranging from years 2010 to 2015 ...
1
vote
0answers
40 views

Unbalanced two-factor repeated measures ANOVA with missing values

For my data set, I need to perform some sort of two factor repeated measures ANOVA. I have one between-subject factor called "Treatment" and one within-subject factor called "Frequency" with 8 levels. ...
1
vote
1answer
95 views

High Recall - Low Precision for unbalanced dataset

I’m currently encountering some problems analysing a tweet dataset with support vector machines. The problem is that I have an unbalanced binary class training set (5:2); which is expected to be ...
1
vote
1answer
42 views

Model Decay of Random Forest, when does it require an update?

I have built a random forest model on a dataset with a large class imbalance, I have attempted to maximize area under the curve when predicting on the test set. I wish to make a suggestion on when the ...
0
votes
0answers
28 views

Unbalanced three-class classification problem

I have three classes which are pretty unbalanced: A, B, and C with 3343, 135 and 1219 observation each respectively. Classes A and C are linearly separable (with ~96% accuracy), while the class B ...
1
vote
0answers
31 views

How to handle with “in class” imbalance in machine learning?

A lot is written about class imbalance in machine learning (for example on this site here). However, how to deal with "intra class" imbalance? Assume I want to classify Bikes v.s. Cars. My training/...
0
votes
0answers
30 views

What techniques can I use to perform feature selection in the context of classification with an highly unbalanced dataset ?

I'm dealing with CTR prediction, which is a classification problem with an highly unbalanced dataset (around 1 positive class for 200 negative class). Most of my features (>90%) are categorical. ...
1
vote
1answer
222 views

Downsampling vs upsampling on the significance of the predictors in logistic regression

I've been trying to build a binary classification model using multivariate logistic regression using the caret package in R. My dataset consists of around 20000 observations from which >99% belongs to ...
2
votes
0answers
16 views

Interplay of Training Class Sizes, Class Weights, Loss function and Decision Threshold

I am facing a two-class classification problem where: There is way more training data in class 1 than in class 0. Classifying a class 0 event as class 1 has a higher loss than classifying a class 1 ...
0
votes
0answers
158 views

Is gradient boosting appropriate for data with low event rates like 1%?

I am trying gradient boosting on a dataset with event rate about 1% using Enterprise miner, but it is failing to produce any output. My question is, since it a decision tree based approach, is it even ...
2
votes
0answers
42 views

mixed effects modelling of unbalanced repeated measures data

I have radio tracking data on 34 animals over a period of up to 26 months. For about 6 animals I have all the data, for 2 others I only have a couple of months, and the rest lie somewhere in between. ...