15 votes
Accepted

Negative prediction values from linear regression in R

You have a linear fit that does not predict well for cars older than ten years. This is because most data points are for cars younger than 10 years old and these will dominate the fitting. If you ...
user avatar
6 votes
Accepted

Logistic Regression on multiple classes (Shouldn't it be only on binary?)

Extending my comment into an answer: There are several natural extensions to handle multiple classes, and two are built into scikit-learn. Multinomial logistic ...
user avatar
6 votes

Negative prediction values from linear regression in R

You didn't constrain the output. Without such a constraint, you allow for any real number to be predicted, including numbers that are ridiculous. For instance, logistic regressions constrain the ...
user avatar
  • 30.8k
3 votes

Choice Between Alternatives in Machine Learning

It's not a machine learning problem and it is a bad idea. First, it is ethically dubious to have black-box software to make career decisions that would potentially influence the future of those ...
user avatar
  • 113k
3 votes
Accepted

Appropriate way to get cross validated performance metrics

I think if the folds are of equal size, then both methods 1 and 2 will give the same mean value (or very similar if the folds are only of approximately the same size). Personally I would tend to use ...
user avatar
2 votes
Accepted

Predictive parametric models and their (unknowable?) coefficients signs

Your first question: "Is there any sense in which because of the counter-intuitive sign on Exercise, a predictive model that includes Exercise is flawed or we'd care less about estimating that ...
user avatar
  • 2,843
2 votes
Accepted

What reasons beyond interpretability are there to use additive models over a complex, multivariate smoother?

There's a problem with multivariate smoothing in high dimensions. Conceptually, a multivariate smooth predicts $f(x_0)$ using data $(x,y)$ where $x$ is 'close' to $x_0$ in every dimension. An ...
user avatar
2 votes

Applications of Dynamic Time Warping (Time Series)

DTW is an algorithm for measuring the distance between two time series. It's an alternative to the Euclidean distance (which is the mean squared distance between the time series at each time step), ...
user avatar
  • 422
2 votes

Features are Relevant for Regression but not necessarily for Classification - what to make of this?

Loosely speaking, I would interpret it to mean that a subset of features are most important for determining the direction (gain/loss), and then the other features come up in determining the magnitude. ...
user avatar
  • 30.8k
1 vote

How to label target dataset based on reduced dimensions of a source dataset?

You can apply the same principal component transformation from the source dataset to the target dataset. This will map your target dataset from the initial N dimensions to the same M dimensions, ...
user avatar
1 vote

Intuition for confidence intervals vs prediction intervals for linear regression

A confidence interval is for the mean of a group of people who have the same input values for your X. If all assumptions are met, 95% of the confidence intervals you calculate will contain the true ...
user avatar
1 vote

Negative prediction values from linear regression in R

[UPDATED] A typical times series interpretation doesn't apply in this case. It's more like a panel data, just to point out. There are more than one value per year. I suspect there are many models of ...
user avatar
1 vote

Forecasting based on few samples

Your data can be plotted as follows: Note: Always plot your data! Especially if you want to forecast. In covid models, a V-shape recovery has been quite frequent. The blue line is your data. The red ...
user avatar
1 vote

Forcing covariates to always be part of a Lasso model

Lasso by default adds a regularization penalty for all the parameters, but nothing prohibits you from penalizing only some of the parameters. Running lasso and "adding back" the zeroed-out ...
user avatar
  • 113k
1 vote
Accepted

When to drop correlated features?

"Significant correlation" would usually mean that you tested a null hypothesis that $\rho=0$. Depending on your sample size, such correlation may still be quite close to zero. Why would you ...
user avatar
  • 113k
1 vote

How to plot the random and best model in Lift and gain charts?

I have never seen these kinds of plots before, but here's the idea. The "random" line is representing the results you would get if you did completely random guessing - so this is the "...
user avatar
1 vote

Is underdispersion problemetic for predictive poisson models?

If you are only interested in the predictions, then a Poisson regression model and a quasiPoisson regression model gives identical predictions, whether there are under- or over-dispersion. The ...
user avatar

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