Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
DanielVandH authored Aug 31, 2024
1 parent d78d287 commit 78c8471
Showing 1 changed file with 2 additions and 2 deletions.
4 changes: 2 additions & 2 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ We consider the problem
```math
y_i = \beta_0 + \beta_1 x_{1i} + \beta_2 x_{2i} + \beta_3 x_{1i}x_{3i} + \beta_4\beta_5 x_{4i} + \varepsilon_i, \quad \varepsilon_i \sim \mathcal N(0, \sigma^2), \quad i=1,2,\ldots,n,
```
and our aim is to estimate $\boldsymbol\theta = (\beta_0, \beta_1, \beta_2, \beta_3, \beta_4, \beta_5 \sigma^2)$. Notice that the parameters $\beta_4$ and $\beta_5$ appear only as a product, so there may be identifiability issues with $\beta_4\beta_5$. To start, we generate some data.
and our aim is to estimate $\boldsymbol\theta = (\sigma, \beta_0, \beta_1, \beta_2, \beta_3, \beta_4, \beta_5)$. Notice that the parameters $\beta_4$ and $\beta_5$ appear only as a product, so there may be identifiability issues with $\beta_4\beta_5$. To start, we generate some data.
```julia
using Random, Distributions, StableRNGs
rng = StableRNG(98871)
Expand Down Expand Up @@ -145,4 +145,4 @@ fig
```
![Profile likelihood plots](profile_likelihood_reparam.svg)

We see that the product $\beta_4\beta_5$ is now identifiable (and $\beta_4$ is still not).
We see that the product $\beta_4\beta_5$ is now identifiable (and $\beta_4$ is still not).

0 comments on commit 78c8471

Please sign in to comment.