14 votes

Success Stories of "Statistics"?

The whole history of statistics is full of them. For example, $t$-tests were born in Guinness brewery as means of optimizing their processes: T-Distribution, also known as Student's t-distribution, ...
11 votes

Do Statistical Binning Algorithms Exist?

The most rational and elegant solution, and best performing in terms of mean squared error of estimates, is to use a method that borrows information across groups: either penalized maximum likelihood ...
user avatar
10 votes

Success Stories of "Statistics"?

The German Tank Problem is a statistical approach to estimating a population size given a sample. The goal is to estimate the total number of items $N$, given a random sample of the population which ...
8 votes
Accepted

Can a variable be linearly independent, but non-linearly dependent?

Yes, you can have zero correlation when you have a nonlinear dependency. For example, $y=x^2$ will have a correlation coefficient of zero, and the linear regression will fit a horizontal line. but ...
user avatar
  • 579
7 votes
Accepted

XGBoost when P>>N

Data is king, so if it works in real life we can't argue with it. Having said that, I agree with you it's bad practice and will usually not end well. I can design a dataset where it would work though. ...
user avatar
  • 579
6 votes

Success Stories of "Statistics"?

To add some medical examples to the excellent cases already cited by others: Richard Doll established the link between smoking and lung cancer. Although a medical doctor, the link was established ...
3 votes

Best way to obtain probabilities and model explanations with imbalanced data

Model training Using log-loss is a good idea because it is a proper scoring rule. Stratified cross validation also sounds like a good idea, but with so few samples (only about 2,500 in the 500,000) I ...
user avatar
1 vote

Choosing among recent references on Machine Learning

I would recommand the new (2022) version of the Probabilistic ML book from Kevin P. Murphy: https://probml.github.io/pml-book/book1.html. I am a huge fan of the approach from the first version: it has ...
1 vote

Best way to obtain probabilities and model explanations with imbalanced data

There is a difference between finding the best model and finding a model that predicts the correct probabilities. In your case using the raw data may lead to the model allways predicting the negative ...
user avatar
  • 579
1 vote
Accepted

interpreting confusion matrix results

To put it slightly differently, your model is able to catch 80.77% of the people who dropped out of workforce, and classified the rest otherwise able to catch only 18.52% of the people who are ...
user avatar
  • 52.5k
1 vote

How deep is the connection between the softmax function in ML and the Boltzmann distribution in thermodynamics?

Here is an academic paper published in the Journal of Statistical Physics by various physicists at MIT: https://dspace.mit.edu/handle/1721.1/135715. The paper discusses the relation of the softmax ...
user avatar
1 vote

How should I train my CNN with a tiny dataset

You should start from a pre-trained model, replace it's output layer with a 3 class classification layer and finetune your model on your images. This is a standard procedure. Here's an example in ...
user avatar
  • 579

Only top scored, non community-wiki answers of a minimum length are eligible