The Stack Overflow podcast is back! Listen to an interview with our new CEO.

Questions tagged [machine-learning]

Machine learning algorithms build a model of the training data. The term "machine learning" is vaguely defined; it includes what is also called statistical learning, reinforcement learning, unsupervised learning, etc. ALWAYS ADD A MORE SPECIFIC TAG.

13,329 questions
Filter by
Sorted by
Tagged with
27 views

performance of a regression model

I am doing random forest regression for multiple features. Now I want to know my model performance. But there is no confusion matrix or accuracy matrix for checking regression model performance. So ...
10 views

Bayesian Optimization Expression

I am reading a paper on Bayesian optimization and came across the following formula: Now my questions are: Question 1: Does the expression inside the redbox evaluate to $p(\mathcal I_{t,0})$? ...
12 views

How is this simplified pearson coefficient derived?

I came across a variation of the Pearson coefficient as seen here,: where x_ti is a target value and x_pi is a predicted value. I've seen the various forms of the r coefficient on wikipedia : https:...
17 views

how to choose the number of patches in sub-region classification before appling CNN?

I'm currently working on image classification, i want to divide the input image into sub-regions(patches) and apply a deep convolutional neural network (CNN) to train and classify each patch ...
19 views

Hard time understanding the parameters of a prediction interval [closed]

I stumpled upon the definition of a prediction interval from https://www.statisticshowto.datasciencecentral.com/prediction-interval/ In my setup I have a type of tree-based regression model with non-...
16 views

can i use logistic regression in discrete feature types

can i use logistic regression in discrete feature types problem. Since recently i knew that the expression for logistic regression is derived from guassian naive bayes and we can use guassian naive ...
10 views

Calculating a baseline probability model for images

I'm a newbie to statistics, so I apologize if this question is trivial. I'm trying to build a distribution that can predict a specific set of images. But first, I need a baseline - so, I decided to ...
10 views

Convexity of Multinomial Logistic Regression

For completely separable training data, is the Multinomial Logistic Regression performance function(Maximum Likelihood) convex? In general, independent of the separability is the Maximum Likelihood ...
8 views

How to implement smogn(smote for regression) in python? [closed]

I have found the paper SMOGN: a Pre-processing Approach for Imbalanced Regression (2017) which gives a github link for code in R. Is there an implementation for smogn in any of the python libraries ...
21 views

Instructive example for using machine learning to predict continuous variable?

Arguably, the MNIST or titanic examples are probably the most common and instructive examples of classification problems solved using machine learning. Are there any 'iconic' examples of machine ...
6 views

consider the outcomes of tossing a K-sided die, is it reasonable to view K as m in “the multinomial formula”? [duplicate]

In mathematics, the multinomial theorem describes how to expand a power of a sum in terms of powers of the terms in that sum. It is the generalization of the binomial theorem from binomials to ...
13 views

If predicted responses has an RMSE/MAE, how often will a prediction fall below the error? [closed]

As the title asks, If I for instance has a regression based model with an error rate, either in terms of RMSE (could be interpreted as a standard deviation?) or MAE, how often will the predicted value ...
28 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 ...
37 views

19 views

Marginal In Bayesian Optimization Expression

I am reading a paper and presentation on batch Bayesian optimization and came across the following formula. Question 1: Does the expression inside the redbox evaluate to $p(\mathcal I_{t,0})$? ...
18 views

Comparing GLM in R with custom SGD [closed]

I am working on a project which requires a costum implementation of an SGD in Scala to solve logistic regression problems. As a baseline for correctness, I have among other things compared the results ...
8 views

Regressor ordering in randomForest [closed]

Consider this example. First, setup the workplace. ...
21 views

Lead Qualification ML Algorithm Guidance

My company has tasked with me with building a ML model to qualify sales leads for our product, and after our brainstorming session, I am not sure how to approach the type of solution they seem to be ...
12 views

Network learns bias during the first iterations if parameter initialization is not good

Andrej Karpathy in his blog post "A Recipe for Training Neural Networks" states that initialization is important for convergence. I get that but when he says: init well. Initialize the final layer ...
108 views

In the example of guess a specified number between 1 and 20 (both inclusive), what is the sample space?

This post is discussing Bayesian reasoning in the context of guess a specified number between 1 and 20 (both inclusive). Consider the following example: I’m thinking of a number between 1 and 20 (...
19 views

Weight initialization in neural networks

Hi I am developing a neural network model using keras. code ...
39 views

How exactly does knowing a variable's probability distribution help you when learning about data?

I am an elementary/wanna-be statistician/data scientist from South Korea. I have been studying a variety of theories of mathematical statistics and different probability distributions. (I apologize ...
5 views

Suggesting multiple labels prediction for multi-class training data

I have a dataset of tickets the machine gets for some machine components failure. The ticket are in text form. For each failure we have around 8-10 diagnosis labels. This tells what the issue might ...
16 views

What is the best way to model a spatialtemporal (3d) problem?

A very common problem in machine learning is that we have time variables. For this we use more statistical approaches like ARIMA or more ML approaches like LSTM. A sophistication of a time series is ...
8 views

Reconstructing face from randomised embedding

It is fairly agreed in literature that from a given face-embedding (that is a vector of features values) it is possible, with a good amount of effort, to reconstruct the original face, (See here for ...
6 views

Error while performing multiclass classification using Gridsearch CV

I am trying to solve a multiclass classification problem using SVC as the base estimator and GridSearchCV to tune my results. Mentioned below is the code and the error being received: ...
5 views

In-Bag Risk Reduction Interpretation - glmboost Variable Importance

I'm currently working wiht a 'glmboost' model, from the 'mboost' R package, and I want to analyze the variable importance. There is a function called 'varimp' that returns the "In-bag risk reduction ...
54 views

Standardise components of an additive model output

I've got a sales forecasting model using the fbprophet library. The model is additive: calculates a base trend and then adds ...
18 views

Statistical significance testing and confidence intervals using RepeatedKFold

How do I get a confidence interval of a model measure significance between models when I do repeated KFold cross validation using ...
7 views

In the context of Generative Classifiers, are the term **priori** and **class conditional density** the same?

this post says There are two main types of models for classification that we consider. The models for classification that we will discuss fall into one of these two categories: Generative ...
8 views

Prove that in such cases, it is possible to ﬁnd an ERM hypothesis for $H_n$ in the unrealizable case in time $O(mnm^{O(n)})$

Let $H_1$ , $H_2$ ,... be a sequence of hypothesis classes for binary classiﬁcation. Assume that there is a learning algorithm that implements the ERM rule in the realizable case such that the ...
11 views

Is using the target value in sample weights target leakage?

If I'm training a regression model, and I want to weight the importance of each sample, is using (a function of) the target as the weight considered target leakage? Does this depend on the particular ...
14 views

(SVMs) Do the specific higher dimensional mappings of attributes not matter when calculating a kernel?

From what I know, one of the strategies employed by an SVM is to increase dimensionality of your data until they are linearly separable. (I guess there's some mathematical proof that your data will ...
92 views

Why is there a bias variance tradeoff? A counterexample

Suppose that $$y=f(x)+\epsilon$$ Where $\epsilon$ has mean $0$ and variance $\sigma^2_e$, independent of $x$. Here is the composition of the mean-squared error into bias and variance: \begin{...
12 views

Sampling and Standardization of data before applying dimentionality reduction?

I'm trying to solve a classification problem with 4 parameters, next_action - binary variable(0/1) total_visits- numerical value days_Since_last_visit - numerical lead_source- categorical variable (5 ...
6 views

Particle Swarm Optimisation Explained

Is anyone able to provide an intuitive explanation of how particle swarm optimisation works? For example how to minimise a function f(x,y,z). Also, does particle swarm optimisation work for multi-...
36 views

It seems that page 32 of “MLaPP” is using notation in a confusing way, I made a little bit enhancement, could someone double check my work?

It seems that page 32 of "Machine Learning: A Probabilistic Perspective by Kevin Patrick Murphy" is using notation in a confusing way. Define the function $F(q) = p(X ≤ q)$. This is called the ...
7 views

Is there a theorem that guarantees an infinitely many number of neural networks that satisfy universal approximation theorem (UAC)?

Given a empricial dataset and its target, Universal Approximation Theorem (UAC) guarantees the existence of at least one neural network that can memorize the target of the dataset. Is there any ...
29 views

Machine Learning Features Recorded At Different Timestamps & Locations

Suppose that you are trying to model the following hypothetical situation - where each of your features is a separate time series, as is the target variable. However the feature time series occur ...
14 views

Is skip-gram model of word embedding actually a multi-class task not a multi-label task, right?

So curious about this question, that I can't describe it in short. Please forgive me. Description： From multiclass and multilabel algorithms, we can get the definition of the multi-class and multi-...
30 views

What is the best way to handle ordinal features having numeric values in python? [closed]

What is the best way to encode ordinal feature? Is it by transforming it using OneHotEncoder so values going from 1 to 7 lets say would become head of new field feature. Or by using StandardScaler() ...
45 views

Should I use traditional machine learning or time series?

I have a debate with my supervisor on whether to use a traditional machine learning model or time-series/Markov chain to solve this problem. We have a set of feature that arranged according to time: ...
26 views

Connection between Stochastic Neighbor Embedding and MDS

In the original SNE paper the authors mention a connection between the SNE objective function and an MDS-like stress function in the regime $\sigma_i \rightarrow \infty$, as follows. When \$\sigma_i^...