1

Heteroskedasticity is relevant in cases in which you calculate a standard error for the estimated coefficients. For instance for a regression model with a single independent variable this would be for the slope coefficient: $$SE(\hat{\beta_1}) = \sigma \left(\frac{1}{\sum_i(x_i - \bar{x})^2}\right)$$ with (see this for more details) $$ \hat{\sigma}^2 = \frac{...


1

It also applies to other methods, i.e. not just linear regression. For example, ANOVA and T-test also depend on homogeneity of variance. One method to check the homogeneity of variance, compatible with the one-way ANOVA, is the Barlett's test.


1

The notation $D_\mathrm{KL}(P \| Q)$ is pretty much standard for the Kullback-Leibler divergence. There is an interesting discussion on Mathematics SE on the reasons for the fairly unusual notation used for divergences. In general, $x \sim G(z, c)$ means "$x$ is a random variable distributed according to $G(z, c)$". In the subscript the author ...


1

Yes - it is $d\times l$. There would be 6 $β$ coefficients for 2*3: $β_1x_1 + β_2x_1^2 + β_3x_2 + β_4x_2^2 + β_5x_3 + β_6x_3^2$ That does not include an intercept or any interaction terms.


1

One option would be look at conditional probability based on percentiles. First, find percentiles based on all the data. For example, 99th percentile is 1432 milliseconds. Then, find percent of a specific user request above that threshold. For example U1 has 50% of requests above 99th percentile. This could be made into a cross-tabs table for easier ...


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