# Tag Info

1

To formalize things a little bit, you are solving a binary classification problem where you have a modest level of class disbalance. It would be helpful to get a more detailed description of what you mean by selecting the best method and model. Given that you have a dedicated test dataset and, as far as I understood, you used cross-validation (by the way, ...

1

My question is when interpreting the categorical covariates, for example, district 1 hazard is 50% more than district 3. Is the increased hazard due to the continuous covariates already in the model, which itself has to be interpreted for each unit of increase in the covariate or the 50% increased hazard from the district is from some other factors not ...

0

Well, first of all note that the regression is equivalent to the one below: $Y = \begin{cases} a\log(x) + b\log(y) + c\log(w) & \text{for} \;\; Z = 0 \\ d + (a+e)\log(x) + (b+f)\log(y) + (c+g)\log(w) & \text{for} \;\; Z=1 \end{cases}$ It's just a neat way of writing two regressions for two cases of categorical variable $Z$. Then, you have a level-log ...

1

"Is the increased hazard due to the continuous covariates already in the model?" has to be answered using interaction terms, i.e., $\texttt{region*saltconcentration}$, $\texttt{region*ADT}$, etc. I would also run models for separate regions, since what's informative regarding a particular set of covariates for one region, may bomb out in another ...

3

Evidence being "correlational" (more often called "observational") means that this was passive observational evidence of a statistical association between things, without any controlled experimentation. For example, consider the claim that "[t]he more people trust science, the more scientifically literate they are." Presumably ...

4

Both the cited finding that, "People who trust the media more are more knowledgeable about politics and the news" and the cited finding that, "The more people trust science, the more scientifically literate they are", are results found in observational data. It is well established that it is difficult to infer causality from ...

3

It means, that changing one of those variables may, as well as may not, lead to changing the other. We can not apriori say which is how much likely. Imagine, that you can force somebody to increase the knowledge about science. If evidence is correlational it may or may not result in changing the trust. If there existed causal evidence, that knowledge causes ...

0

What does the bolded phrase mean? I agree with you that it means the evidence isn't causal. In other words, no causal inference was tested so they only note the correlation in the survey data. Does it mean that even if the evidence isn't causal, the next phrase still makes sense? Yes. Even if the relationship isn't causal, only an experimental ...

2

Trust in Science -> Scientific Literacy is observationally equivalent with Scientific Literacy -> Trust in Science. Similarly, Education -> Trust in Science & Education -> Scientific Literacy is also observationally equivalent with Trust in Science -> Scientific Literacy. So when they say "even if this evidence remains correlational&...

1

if you want to get the z-score, try installing the package "FSA" instead. See the image of the code below. Cheers.

2

A model with a logit link function assumes a linear relationship between the logit-transformed variable (log-odds when we speak about logit transformed probabilities) and the predictors. (Note that this assumption of linearity might be false and it can lead to the piranha problem when we extrapolate too far or with too many combined multiple effects) Below ...

4

Another fun thread! @COOLSerdash is one of my favourites on this forum! Tom, when you model an outcome variable expressed as a proportion, you have to be a bit careful with your modelling, as I'll explain below. In statistics, we tend to think of a proportion as either discrete or continuous. How you model discrete proportions is generally different from how ...

0

Maximum likelihood estimation works by trying to maximize the likelihood. As the log function is strictly increasing, maximizing the log-likelihood will maximize the likelihood. We do this as the likelihood is a product of very small numbers and tends to underflow on computers rather quickly. The log-likelihood is the summation of negative numbers, which ...

0

The thread has long been inactive, but I'd like to add one thing to Maarten Buis's already perfect answer. The logit model is $\log( p(X)/(1-p(X)) ) = X\beta$, where $p(X) = P(Y=1|X)$. When $X$ increases from $x$ ($\mathit{before}$, say) to $x+\Delta x$ ($\mathit{after}$, say), the log-odds increases by $\Delta x\beta$: $$\log (\mathit{odds}_{after}) - \log (... 0 To answer your question from a totally uncorrelated perspective, it's instructive to note that in financial engineering and economics, there needs to be some kind of uncertainty around cost or budget estimates, since a single prediction based on the best features won't be sufficient. Even if you employ ML and determine which features are the most ... 1 Machine learning explainability is an area of active research, quite popular in recent years. This question cannot be answered briefly, because there are many research papers, tutorials, and even books on this subject. It would be an impossible task to summarize them in few sentences. I would encourage you to check the freely available online book ... 0 I will answer my own question given that I could solve the issue, and the problem was not related to coefficient interpretation. In short, removing the intercept alters the meaning of the logistic regression coefficients. So, best to keep it even if it is not significant. The main issue here was the data treatment I was doing prior to training the algorithm ... 4 Having read the blog post, I think the author is saying that we shouldn't use randomness in models of the real world because the real world is not random, since everything (such as a coin flip) actually has a cause. This makes probability theory the science of last resort. Only after truly exhausting your ability to investigate causal factors and processes ... 4 I think the issue with the arguments raised in the question is the naive realist philosophy of models apparently behind it. If we model an experiment in a frequentist manner, what we do is that, when using the model, we treat the experiment as if it would be infinitely repeatable, with random outcomes the relative frequency of which stabilises for a growing ... 4 As to the first critique, it could be a critique of any and all branches of the sciences. There are no perfectly repeatable experiments. It isn't really possible to completely control any experiment. A meteor could strike the location of the experiment, for example. Also, the ability to repeat an experiment is irrelevant. Most Frequentist inferences are ... 0 In your model,$$E[\ln y \vert x, z]=\alpha + \beta \cdot x + \gamma \cdot z.$$Here z stands for other covariates. Taking the derivative of that with respect to x, you get$$\frac{1}{y} \cdot \frac{\partial y}{\partial x} = \beta \approx \frac{\frac{\Delta y}{y}}{\Delta x}.$$This means that$$ \beta \approx \frac{\frac{\Delta y}{y}}{\Delta x}, so ...

5

Statistical significance requires that the hypotheses are prespecified. There's no way on Earth the authors prespecified each coefficient in that huge table as a hypothesis. If they did, they ought to adjust for multiple comparisons, effectively throwing the results in the trash. You can calculate the significance test with the coefficient and standard error ...

2

This is a technique used to understand patterns of missing values in a dataset. Given two variables $X$ and $Y$ (which are in effect equal-length arrays of values, which are either references to computer objects -- of any type -- or a special code to representing a missing value), we can quantify their patterns of missing values using some kind of "...

0

Association of null value in columns is how much it correlates with other column For nullinity matrix correlation calculation Refer, How to calculate nullity correlation matrix? Ex: In house price calculation, we will gather swimming pool and gardening data only if it is individual home. If it is appartment, we will gather the data. In this case it is filled ...

0

The left branch is a yes in R, see also here: https://www.datacamp.com/community/tutorials/decision-trees-R There is an R output of your model that gives you the importance of your features. In case you want to calculate them manually: Here is a like that shows you with an example, how you can calculate the feature importance (known from sklearn) for all ...

0

Your models are nested. In the R formula language, : means interaction, while * means "expand into a sum plus the interaction". So the term income * gender expands into income + gender + income:gender, in effect including the possibility that slopes of income differs between genders, while the simpler model without the interaction assumes that ...

0

Model 2 is nested within Model 1, as setting the income:gender parameter in Model 1 equal to zero results in Model 2. Thus, Model 2 is the reduced model. Model 1 $\hat{y} = \hat{\beta}_0 + \hat{\beta}_1 x_{income} + \hat{\beta}_2x_{gender} + \hat{\beta}_3 x_{income}x_{gender}$ Model 2 \$ \hat{y} = \hat{\beta}_0 + \hat{\beta}_1 x_{income} + \hat{\beta}_2x_{...

Top 50 recent answers are included