Skip to main content
Commonmark migration
Source Link

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2](1) (2) $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'\Omega r}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$

Edit: I removed the externally studentized Residuals given that I use a value of $\sigma$ that is not based on the residuals.

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'\Omega r}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$

Edit: I removed the externally studentized Residuals given that I use a value of $\sigma$ that is not based on the residuals.

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. (1) (2) $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'\Omega r}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$

Edit: I removed the externally studentized Residuals given that I use a value of $\sigma$ that is not based on the residuals.

or another alternative ?

Notice removed Canonical answer required by John Tokka Tacos
Bounty Ended with cherub's answer chosen by John Tokka Tacos
deleted 50 characters in body
Source Link

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'Wr}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$$$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'\Omega r}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$ Externally

Edit: I removed the externally studentized Residuals $\sim t(\nu)$, sogiven that I use a value of $Var(r_{i,IntStud}) \approx \frac{\nu}{\nu-2}$ $$ r_{i,IntStud} = \frac{r_i}{\sqrt{\sigma_{(i)}^2 (\frac{1}{w_{i}} - H_{ii})}} $$$\sigma$ that is not based on the residuals.

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'Wr}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$ Externally studentized Residuals $\sim t(\nu)$, so $Var(r_{i,IntStud}) \approx \frac{\nu}{\nu-2}$ $$ r_{i,IntStud} = \frac{r_i}{\sqrt{\sigma_{(i)}^2 (\frac{1}{w_{i}} - H_{ii})}} $$

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'\Omega r}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$

Edit: I removed the externally studentized Residuals given that I use a value of $\sigma$ that is not based on the residuals.

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

added 212 characters in body
Source Link

Given a linear model whereWe want to estimate $Y=X\beta+e$$\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $e\sim N(0,\Omega^{-1})\sim N(0,\sigma^2W^{-1})$$x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega=\frac{1}{\sigma_i^2}$$\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'Wr}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma}$$$$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$ Externally studentized Residuals $\sim t(\nu)$, so $Var(r_{i,IntStud}) \approx \frac{\nu}{\nu-2}$ $$ r_{i,IntStud} = \frac{r_i}{\sqrt{\sigma_{(i)}^2 (\frac{1}{w_{i}} - H_{ii})}} $$

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

Given a linear model where $Y=X\beta+e$ and $e\sim N(0,\Omega^{-1})\sim N(0,\sigma^2W^{-1})$, where $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point. We measure a total of $n$ points

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'Wr}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$ Externally studentized Residuals $\sim t(\nu)$, so $Var(r_{i,IntStud}) \approx \frac{\nu}{\nu-2}$ $$ r_{i,IntStud} = \frac{r_i}{\sqrt{\sigma_{(i)}^2 (\frac{1}{w_{i}} - H_{ii})}} $$

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

We want to estimate $\beta$ for $$ y = x\beta + \epsilon $$ where $y$ and $x$ are $n\times 1$ vector and $\epsilon$ is not i.i.d, but $\epsilon \sim N(0, \sigma^2\Omega)$, where $\Omega$ and $W$ are $n \times n$ nonsingular, positive definite and symmetric matrix with $W_{ii}=w_i=\frac{\sigma^2}{\sigma_i^2}$ and $\Omega_{ii}=\frac{1}{\sigma_i^2}$ .

Assume that, thanks to a lot of repeated measurements, we know the underlying measurement uncertainties $\sigma_i$ of my response variable in the i-th measured point, so $\Omega$ and $W$ are known. We measure a total of $n$ points.

Given Heteroskedasticity we use weigthed least squares.

The residual analysis yields good results, showing that the residuals are independent and normally distributed, and that the weights enable studentized residuals with constant variance.

Now what is the best way to assess the goodness of fit ?

1.) Reduced chi-square: $\chi_{red}^2$ should be close to 1. [(1)][1] [(2)][2] $$\chi_{red}^2 = \frac{\chi^2}{\nu} = \frac{r'\Omega r}{\nu} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

N.B.: This corresponds to a comparison of the unbiased estimate of error variance $\hat{\sigma}^2$ and the known mean measurement uncertainty $\sigma^2$. $$ \frac{\hat{\sigma}^2}{\sigma^2} = \frac{r'Wr}{\nu} \cdot \frac{1}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot w_i}{\sigma^2} = \frac{ \frac{1}{\nu} \sum_i^n r_i^2 \cdot \frac{\sigma^2}{\sigma_i^2}}{\sigma^2} = \frac{1}{\nu} \cdot \sum_i^n\frac{ r_i^2}{\sigma_i^2} $$

or

2.) Evaluation of the variance of the standardized/studentized Residuals, which should be close to 1. Note that the value for $\sigma$ would be the one geven by the prior repeated measurements and not the MSE, where:

Standardized Residuals $\sim \mathcal{N}(0,\,1)$, so $Var(r_{i,Stand}) \approx 1$ $$r_{i,Stand} = \frac{r_i}{\sigma_i}$$ Internally studentized Residuals: $$ r_{i,ExtStud} = \frac {r_i}{var(r_i)} = \frac{r_i}{\sqrt{\sigma^2 (\frac{1}{w_{i}} - H_{ii})}}$$ Externally studentized Residuals $\sim t(\nu)$, so $Var(r_{i,IntStud}) \approx \frac{\nu}{\nu-2}$ $$ r_{i,IntStud} = \frac{r_i}{\sqrt{\sigma_{(i)}^2 (\frac{1}{w_{i}} - H_{ii})}} $$

or another alternative ? [1]: https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic [2]: https://arxiv.org/abs/1012.3754

Tweeted twitter.com/StackStats/status/1262171108071047173
added 1176 characters in body
Source Link
Loading
Notice added Canonical answer required by John Tokka Tacos
Bounty Started worth 150 reputation by John Tokka Tacos
Source Link
Loading