Regression error - "NUM!"

[AudreyJ]

Hi,

Can anyone tell me why I might be getting "NUM!" as P-Values in my output?
I'm not sure what that might mean. I deleted a few fields of data to increase
my R Square and all of a sudden I'm getting "NUM!" as several of my P-Values.

Any help would be greatly appreciated!

AudreyJ

 

[headly]

When you say you deleted some fields, do you mean rows or columns?
What type of data are you regressing; dates against numbers? if so, is all
data numeric data type; does working with a smaller data set produce the same
result? You might try to copy and paste special values

 

[AudreyJ]

Hi Headly!

When I say I deleted fields I mean I deleted 4 rows from about 50 (the ones
that had the highest residuals). My results are similar to the results I had
before except I have a higher R Square after deleating the rows and I have
the NUM! error as P-Values. I tried copy paste special values to my data, I
still get the same errors. All of the data I'm regressing is numbers. And
actually I pasted the first row of data below. It's a little messy, but
hopefully you can get the idea...

9/29/6 Midspread (bps) Expected Loss (bps) Multiperil USWD (Base case
CAEQ) EurWD Other Perils Notional ($mm) Indemnity (Base case Parametric
Index/ Pure Parametric/ Modeled Loss) Industry Index MITT
575 194 0 0 0 1 250 0 0 0

Thank you so much for your response,

Audrey

 

[Jerry W. Lewis]

P-value is
=TDIST(t_Stat, df[Residual], 2)
What are the values of "t Stat" and its corresponding "Standard Error" as
well as "df" and "SS" for Residual?

Jerry

 

[Tushar Mehta]

How many degrees of freedom do you have for the residuals? Is it zero?

--
Regards,

Tushar Mehta
www.tushar-mehta.com
Excel, PowerPoint, and VBA add-ins, tutorials
Custom MS Office productivity solutions

 

[AudreyJ]

Hi Jerry,

I tried to paste what you asked about in a way that is clear below, I hope
it comes out okay. I'm not sure exactly how the formula you gave for Pvalue
works. Does the below make any sense to you?

Thank you for your response,

Audrey

Coefficients Standard Error t Stat P-value
Intercept 1,189.0 -
65,535.0 #NUM!
Expected Loss (bps) 1.7 0.3 4.9 0.0

df SS
Regression 9.0 2,790,427.3
Residual 49.0 561,151.7

P-value is
=TDIST(t_Stat, df[Residual], 2)
What are the values of "t Stat" and its corresponding "Standard Error" as
well as "df" and "SS" for Residual?

Jerry

Hi Headly!

When I say I deleted fields I mean I deleted 4 rows from about 50 (the ones
that had the highest residuals). My results are similar to the results I had
before except I have a higher R Square after deleating the rows and I have
the NUM! error as P-Values. I tried copy paste special values to my data, I
still get the same errors. All of the data I'm regressing is numbers. And
actually I pasted the first row of data below. It's a little messy, but
hopefully you can get the idea...

9/29/6 Midspread (bps) Expected Loss (bps) Multiperil USWD (Base case
CAEQ) EurWD Other Perils Notional ($mm) Indemnity (Base case Parametric
Index/ Pure Parametric/ Modeled Loss) Industry Index MITT
575 194 0 0 0 1 250 0 0 0

Thank you so much for your response,

Audrey

 

[AudreyJ]

Hi Jerry,

I tried to paste the values that you asked about in a way that is readable
below. I'm not sure exactly how the formula you sent me for PValue works. I
don't have much experiance with this type of thing. Do you see anything in
the below?

Thank you for your response,

Audrey
df SS MS F
Regression 9.0 2,790,427.3 310,047.5 27.1
Residual 49.0 561,151.7 11,452.1
Total 58.0 3,351,579.0

Coefficients Standard Error t Stat P-value
Intercept 1,189.0 - 65,535.0 #NUM!
Expected Loss (bps) 1.7 0.3 4.9 0.0

P-value is
=TDIST(t_Stat, df[Residual], 2)
What are the values of "t Stat" and its corresponding "Standard Error" as
well as "df" and "SS" for Residual?

Jerry

Hi Headly!

When I say I deleted fields I mean I deleted 4 rows from about 50 (the ones
that had the highest residuals). My results are similar to the results I had
before except I have a higher R Square after deleating the rows and I have
the NUM! error as P-Values. I tried copy paste special values to my data, I
still get the same errors. All of the data I'm regressing is numbers. And
actually I pasted the first row of data below. It's a little messy, but
hopefully you can get the idea...

9/29/6 Midspread (bps) Expected Loss (bps) Multiperil USWD (Base case
CAEQ) EurWD Other Perils Notional ($mm) Indemnity (Base case Parametric
Index/ Pure Parametric/ Modeled Loss) Industry Index MITT
575 194 0 0 0 1 250 0 0 0

Thank you so much for your response,

Audrey

 

[AudreyJ]

Hi Jerry,

I tried to paste the values that you asked about in a way that is readable
below. I'm not sure exactly how the formula you sent me for PValue works. I
don't have much experience with this type of thing. Do you see anything in
the below?

Thank you for your response,

Audrey
df SS MS F
Regression 9.0 2,790,427.3 310,047.5 27.1
Residual 49.0 561,151.7 11,452.1
Total 58.0 3,351,579.0

Coefficients Standard Error t Stat P-value
Intercept 1,189.0 - 65,535.0 #NUM!
Expected Loss (bps) 1.7 0.3 4.9 0.0

P-value is
=TDIST(t_Stat, df[Residual], 2)
What are the values of "t Stat" and its corresponding "Standard Error" as
well as "df" and "SS" for Residual?

Jerry

Hi Headly!

When I say I deleted fields I mean I deleted 4 rows from about 50 (the ones
that had the highest residuals). My results are similar to the results I had
before except I have a higher R Square after deleating the rows and I have
the NUM! error as P-Values. I tried copy paste special values to my data, I
still get the same errors. All of the data I'm regressing is numbers. And
actually I pasted the first row of data below. It's a little messy, but
hopefully you can get the idea...

9/29/6 Midspread (bps) Expected Loss (bps) Multiperil USWD (Base case
CAEQ) EurWD Other Perils Notional ($mm) Indemnity (Base case Parametric
Index/ Pure Parametric/ Modeled Loss) Industry Index MITT
575 194 0 0 0 1 250 0 0 0

Thank you so much for your response,

Audrey

 

[Jerry W. Lewis]

The standard error for the intercept is reported as (essentially) zero.
Therefore the t statistic is (essentially) infinite, with a p-value of
(essentially) zero.

The #NUM! for p-value appears to be an intervention of the analysis ToolPak,
since the TDIST function does not appear to return #NUM! in this range.

Jerry

Hi Jerry,

I tried to paste what you asked about in a way that is clear below, I hope
it comes out okay. I'm not sure exactly how the formula you gave for Pvalue
works. Does the below make any sense to you?

Thank you for your response,

Audrey

Coefficients Standard Error t Stat P-value
Intercept 1,189.0 -
65,535.0 #NUM!
Expected Loss (bps) 1.7 0.3 4.9 0.0

df SS
Regression 9.0 2,790,427.3
Residual 49.0 561,151.7

P-value is
=TDIST(t_Stat, df[Residual], 2)
What are the values of "t Stat" and its corresponding "Standard Error" as
well as "df" and "SS" for Residual?

Jerry

Hi Headly!

When I say I deleted fields I mean I deleted 4 rows from about 50 (the ones
that had the highest residuals). My results are similar to the results I had
before except I have a higher R Square after deleating the rows and I have
the NUM! error as P-Values. I tried copy paste special values to my data, I
still get the same errors. All of the data I'm regressing is numbers. And
actually I pasted the first row of data below. It's a little messy, but
hopefully you can get the idea...

9/29/6 Midspread (bps) Expected Loss (bps) Multiperil USWD (Base case
CAEQ) EurWD Other Perils Notional ($mm) Indemnity (Base case Parametric
Index/ Pure Parametric/ Modeled Loss) Industry Index MITT
575 194 0 0 0 1 250 0 0 0

Thank you so much for your response,

Audrey


:

When you say you deleted some fields, do you mean rows or columns?
What type of data are you regressing; dates against numbers? if so, is all
data numeric data type; does working with a smaller data set produce the same
result? You might try to copy and paste special values

:

Hi,

Can anyone tell me why I might be getting "NUM!" as P-Values in my output?
I'm not sure what that might mean. I deleted a few fields of data to increase
my R Square and all of a sudden I'm getting "NUM!" as several of my P-Values.

Any help would be greatly appreciated!

AudreyJ

 

[AudreyJ]

Hi Tushar,

Freedom for the residuals is actually up to about 30%.

- Greta