daviddoji/summerschool_simtech_2023
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update Exercises 2 and 3 (Multiple Regression).

Browse Source
This commit is contained in:
Marco Oesting
2023-10-09 17:48:50 +02:00
parent b265c7afe0
commit a9b065ab5f
1 changed files with 19 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
material/3_wed/regression/MultipleRegressionBasics.qmd
View File

@@ -100,9 +100,10 @@ normal distribution with mean 0.0 and standard deviation 2.0, and sample
## Task 1 ## Task 1
1. Generate $n=20$ covariates $\mathbf{x}$ randomly. 1. Generate $n=20$ covariates $\mathbf{x}$ randomly.
2. Given these covariates and true parameters $\beta_0=-3$, $\beta_1=2$ 2. Given these covariates and the true parameters $\beta_0=-3$,
and $\sigma=0.5$, simulate responses from a linear model and $\beta_1=2$ and $\sigma=0.5$, simulate responses from a linear model
estimate the coefficients $\beta_0$ and $\beta_1$. (with normally distributed errors) and estimate the coefficients
$\beta_0$ and $\beta_1$.
3. Play with different choices of the parameters above (including the 3. Play with different choices of the parameters above (including the
sample size $n$) to see the effects on the parameter estimates and sample size $n$) to see the effects on the parameter estimates and
the $p$-values. the $p$-values.
@@ -189,10 +190,18 @@ regression model, but we provide explicit formulas now:
p\text{-value} = \mathbb{P}(|T| > t_i), \quad \text{where } T \sim t_{n-p} p\text{-value} = \mathbb{P}(|T| > t_i), \quad \text{where } T \sim t_{n-p}
$$ $$
**Task 2**: Implement functions that estimate the $\beta$-parameters, ::: {.callout-caution collapse="false"}
the corresponding standard errors and the $t$-statistics. Test your
functions with the \`\`\`tree''' data set and try to reproduce the ## Task 2
1. Implement functions that estimate the $\beta$-parameters,
the corresponding standard errors and the $t$-statistics.
2. Test your functions with the `tree' data set and try to reproduce the
output above. output above.
:::
Which model is the best? For linear models, one often uses the $R^2$ characteristic.
Roughly speaking, it gives the percentage (between 0 and 1) of the variance that can be explained by the linear model.
``` julia ``` julia
r2(linmod1) r2(linmod1)
@@ -305,10 +314,11 @@ model = glm(@formula(participation ~ age^2),
SwissLabor, Binomial(), ProbitLink()) SwissLabor, Binomial(), ProbitLink())
``` ```
::: callout-task ::: {.callout-caution collapse="false"}
**Task 3**:
##Task 3:
1. Reproduce the results of our data analysis of the `tree` data set 1. Reproduce the results of our data analysis of the `tree` data set
using a generalized linear model with normal distribution family. using a generalized linear model with normal distribution family.
2. Generate 2. Generate $n=20$ random covariates $\mathbf{x}$ and Poisson-distributed counting data with parameters $\beta_0 + \beta_1 x_i$. Re-estimate the parameters by a generalized linear model.
::: :::