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.

Filter by
Sorted by
Tagged with
-1
votes
0answers
30 views

LSTM - stock prediction with stock prices + financial news

the problem is, I have financial news data (used sentiment analyser on it) and turned it into dataframe (neg,pos,neutre) and I have stock prices. both are in the form of time series. I want to use ...
0
votes
1answer
21 views

why increasing diversity in ensemble method cause lower variance?

I am reading hands on machine learning . Bootstrapping introduces a bit more diversity in the subsets that each predictor is trained on, so bagging ends up with a slightly higher bias than pasting, ...
0
votes
0answers
9 views

Does different data distribution of training and testing data cause overfitting?

Let's assume that I'm developing a classification model for the product of my company but there's a problem. The problem is the data from my company is not enough to develop the model since my company ...
0
votes
1answer
27 views

Model Overfitting When Employing Upsampling/DownSampling

I have a binary classification problem with a large class imbalance ( 1/100 ). I am getting fair results using ensemble modeling. I understand that one technique that could improve results is ...
0
votes
0answers
6 views

Order of Magnitude on Very Flexible Model

I am taking an online class that rarely has an instructor. Upon reading the lab note of the class, I have a question. We are doing Polynomial Regression and Splines which are very flexible models. In ...
1
vote
1answer
32 views

Q Learning Function Approximation

When I have the following function $$Q(s,a;w) = w_1f_1(s,a)+\cdots +w_nf_n(s,a)$$ in reinforcement learning. $f(s,a)$ is the feature vector. How is $f(s,a)$ defined? When I have state $s_0$ and two ...
0
votes
0answers
36 views

Multiple Regression Coefficients

In section 3.2.3 of Elements of Statistical Learning (Link), there's this statement on multiple regression coefficients on Page 54 we have shown that the $j^{th}$ multiple regression coefficient is ...
6
votes
2answers
80 views

What to do when your training and testing data have different distributions

I am training a XGBoost regression model for predicting number of applications and the range of the target variable in train and test data set is different. For e.g: In Train data : Minimum ...
0
votes
0answers
5 views

Is spare encoding a special case of spare autoencoding with ignoring non linear activation

we know that Sparse encoding is to minimize the objective function: $$\sum\limits_{n=1}^N\Big(||x_n - Az_n||^2 + \lambda\rho(z_n)\Big).$$ Here $A = [a_1,\cdots,a_M]...
0
votes
0answers
6 views

Is supervised learning harder under multiple-labels than when labels are mutually exclusive?

It is common to encounter problems that involve some form of multi-class supervised learning. Within this category, there are two possibilities. One that the classes are mutually exclusive (...
0
votes
0answers
23 views

How does AlphaZero guarantee it could make consistent improvement?

I know the detail of AlphaZero. And in detail, I know it is improving by "policy iteration" mechanism. I found an answer that prove it can finally converge to optimal. But... Is it still ...
4
votes
1answer
56 views

Why the Lasso cost function isn't differentiable at $\theta_i=0$ and what is the effect of $g$?

I'm reading Hands-On Machine Learning by Aurélien Géron. The author states that the Lasso cost function isn't differentiable at $\theta_i=0$ so we use a subgradient vector $g$ instead of the gradient ...
0
votes
0answers
12 views

Log Log Regression Response value, when logged predictor is zero [duplicate]

In a log-log linear model in R, How can one get the contribution amount (not the coefficient) of each of the explanatory variables (independent variables) (i.e. x1, x2, x3), when log-log form is ...
1
vote
0answers
43 views

For a nonlinear regression task, is either Maximum Likelihood Estimation or Least Squares easier to learn a neural network model with?

I have data (x,y) and I want to create a model f(x) that will best approximate y. Let's ...
3
votes
1answer
32 views

Can MAPE values change after inverse tranform of target?

I have tranformed my Target variable as following Maxmin transformation and used an engine to get prediction for a time series data. The target transformation is done using the following formula in ...
0
votes
0answers
8 views

Is there a need to standardise training and test sets separately for binary classification problems? [duplicate]

When setting up an ML framework for binary classification do we need to standardize our training and test sets separately? This answer claims to standardize separately (although never states the ...
2
votes
1answer
20 views

How to prepare a 2x2 Confusion matrix for binary classifier, without code

Problem statement: Evaluate a binary classifier. There are 50 positive outcomes in the test data, and 100 observations. Using a 50% threshold, the classifier predicts 40 positive outcomes, of which 10 ...
1
vote
1answer
17 views

Data normalization of test data in machine learning

I have 117 samples which I used to select and train a model. What I did: 1) pre-processed the 117 samples (normalization, statistics, etc); 2) created 4 folds (random split); 3) performed a nested-...
1
vote
1answer
31 views

Appropriate choice of F1 score

We can compute the F-1 score in the following two ways. $F_{1_{PRE, REC}} = 2 * (PRE * REC) / (PRE + REC)$ $F_{1_{TP, FP, FN}} = (2 * TP) / (2 * TP + FP + FN)$ Both computes F1 score, but which one ...
1
vote
0answers
18 views

Why is the convergence rate faster for this given approach?

LALR: Theoretical and Experimental validation of Lipschitz Adaptive Learning Rate in Regression and Neural Networks This is a paper that suggests using an adaptive learning rate approach for various ...
2
votes
1answer
42 views

Anomaly detection and explanation

I was given a dataset with 500 features which, after one-hot encoding, looks like this: Class = 1 means "anomaly", class = 0 is "normal". So basically my task is simple ML ...
1
vote
0answers
16 views

How to model a new short term beaviour in time series using ML approaches?

I had been modeling time series which contain many SKUs , due to corona there was a short-term drift in seasonality. I want to know how can i deal with this problem. Usually, sales are high in April, ...
2
votes
0answers
8 views

Gradient Boosting algorithm

What is beta for? Why not just fit h_m (get a_m) without this beta? since h_m is an estimator for pseudo-residuals
0
votes
2answers
26 views

Binary Classification problem without using Sigmoid?

I'm have a network where I am using BCE for loss and a sigmoid layer as the last layer's activation function. If I wanted to remove sigmoid, and declare negatives to be one class and positives to be ...
0
votes
0answers
5 views

Designing a Specialized Domain Text Summarizer

I am trying to develop an extractive summarizer for the medical domain. Here I am trying to tag the medical domain entities and classify them. I am trying to give a small example. Mr.XYZ has developed ...
0
votes
0answers
17 views

Posterior predictive of normal normal-mean conjugacy

I want to compute: $$p(x | X) = \int p(x | \mu , \Sigma) p(\mu | X) = \int \mathcal{N}(x | \mu , \Sigma) \mathcal{N}(\mu | \mu_N , \Sigma_N ) d\mu$$ Actually, this is the posterior predictive of the ...
0
votes
3answers
148 views

Is there a “canonical” probabilistic version of the step function?

Step and switch-like functions can be thought of as deterministic switches at some threshold, so smooth sigmoidal-like functions describe that switch with uncertainty around that value. The machine ...
3
votes
0answers
49 views

Xgboost: does data need to be standardized when using shrinkage (parameters lambda, gamma)?

In this similar question about the implications of standardizing the features of data, the answer is that it is not important. However, (as pointed out in the comments on that post) I am interested in ...
0
votes
0answers
25 views

Interpretation of decision tree fancyrpartplot

I constructed the below pruned tree after 10-fold cross validation. I am not sure how to interpet the tree. For example in the root node before the first split I know we have 100% of the data. So then,...
0
votes
0answers
22 views

Why are Deep CNN's mesh dependent?(Fourier Neural Operators article)

I am reading the article named "Fourier Neural Operator for Parametric Partial Differential Equations" by Zongyi Li et al. In the very first page, there is a paragragh about Finite ...
0
votes
0answers
14 views

Major discrepancy of latent variable in the Gaussian Mixture Model/Expectation and Maximization literature

I have read a couple of references on the interpretation of a latent variable in the GMM/EM literature and I found a massive discrepancy between the authors so much so I now have no idea how GMM/EM ...
0
votes
0answers
17 views

Huge AUC difference between using predictions from xgboost.train and XGBClassifier

I'm using two different versions of XGBoost modeling, and seeing that the two versions are producing vastly different AUC results. As far as I know, the XGBClassifier.fit() method should be using the ...
0
votes
0answers
15 views

Repeated measures on individuals with the same outcome

I've attempted to tackle the following problem: When a customer has registered on our site, we'd like to predict if he/she will eventually leave us. I have 3 days' worth of data, that captures ...
0
votes
0answers
16 views

Why using point estimates instead of integrating out the unknown?

I was just wondering why you often use point estimators like MAP and MLE when you have to calculate the posterior distribution for them anyway? Is it because you don't have to calculate the evidence ...
0
votes
0answers
12 views

joint two-stage estimation with ML

I'm working with data of around 15 variables and half a million observations. To avoid selection bias, I'm trying to incorporate a two-stage joint estimation. I've seen this performed as a censored ...
1
vote
1answer
168 views

Bounding the uniform deviation of the empirical risk from the risk over a finite function class

I am having difficulty interpreting the following theorem from here as a probability statement: Theorem. For all $\delta$ such that $0 < \delta < 1/2$, with proability at least $1 - \delta$ the ...
0
votes
1answer
19 views

How to show a significant difference between two prediction populations obtained by different ML methods?

My goal is to compare two diferent ML methods for a prediction problem. I ve run both methods on a simulated data set with a known true value a 1000 times and obtained a prediction distribution as in ...
2
votes
0answers
39 views

How to use neural networks to learn a probability density function?

For example, if we want to learn the pdf or a normal distribution, we can let the network out the mean and variance parameters of the pdf function. However, for other general/unknown distribution, is ...
0
votes
0answers
22 views

Cross entropy error: Poor modelling giving too much weight to unlikely events

I was reading this paper. link (page 5) In this paper, there is a statement that goes like this: To begin, cross entropy error is just one among many possible distance measures between probability ...
6
votes
2answers
115 views

When will $\text{Variance}=\text{Bias}^2$ hold for the optimal model?

Let us consider the bias-variance decomposition in the context of model selection. The picture below suggests the optimal model (the one minimizing the expected squared prediction error) will have $\...
0
votes
0answers
21 views

How to interpret the output of cross-validation for SVR

I wrote this code to run a SVR with cross validation: ...
3
votes
0answers
28 views

Question about minimising empirical loss by gradient descent

Say we wanted to learn $f_{\theta}(\pmb{x})=y$, with a loss function $L(f_{\theta}(\pmb{x}),y)$. We often want to choose $\theta$ which minimises the empirical loss, as the exact loss isn't available ...
1
vote
0answers
32 views

Are SVM generalization bounds valid if the kernel is learned on a different dataset?

Suppose I have a training dataset with a binary label, and I do an 80/20 split on it. On the first 80% of the data, I train a deep learning embedding model that maps my data unto some higher ...
0
votes
0answers
32 views

Combinatorial symmetric CV vs Combinatorial purged CV

Reading "Advances in Financial Machine Learning", and the author proposes 2 methods of CV: "combinatorial symmetric cross validation" (11.6) and "combinatorial purged cross ...
1
vote
0answers
9 views

Partitioning data set for training, testing, and production deployment of machine learning churn classifier

I am new to machine learning. I am working on churn prediction for a customer. I am wondering how best to partition the data for training/test/production deployment. My thinking is that churning ...
0
votes
0answers
10 views

How can i predict the chance of something happening at a specific time and location using historical data?

I have a list of around 1.5k police checkpoints, it includes their location (coordinates) time and I combined that dataset with a weather dataset. My goal is to predict the chance of a police ...
29
votes
5answers
3k views

How can you account for COVID-19 in your models?

How are you dealing with the coronavirus "event" in your machine learning models? Let's say you used to predict the number of sales each month. The virus affected your results last year and ...
0
votes
0answers
30 views

How to train a model, that finds correlation between items, when the data contains many unique identifiers?

I have 3 years of historical data for 40k items that looks like this. Date ID Price Qty. Sold (Target) 2018-01-01 1 1.5 30 2018-01-01 2 2.0 35 ... 2018-01-01 40000 1.0 120 2018-01-02 1 1.5 30 ...
0
votes
0answers
14 views

Is the test set in Combinatorial Purged Cross-Validation in-sample or out-of-sample?

I am trying to understand the Combinatorial Purged Cross-Validation (CPCV) method of Marcos Lopez de Prado's "Advances in Financial Machine Learning" book. There are a few things that I do ...
1
vote
1answer
21 views

Can nested cross validation be considered bootstrapping?

I am using nested cross validation with 5 inner and outer folds. Each of the folds are created using stratified shuffle splits from scikit-learn. Because I am using, can this be considered ...

1
4 5
6
7 8
338