Statistics Help!!!! Urgent! (1 Viewer)

[UTS101]

Hi guys, I have been stuck on a question for a number of hours.... the question comes with the following data set on a shops advertising spending, this is the data set:

Sales DirectNewspaperTelevision
15.98 1.05 1.35 1.55
21.59 1.29 2.25 1.97
25.47 0.96 1.79 1.64
16.02 1.51 1.73 1.73
16.58 1.30 1.71 1.61
20.10 1.06 1.61 1.82
17.49 0.86 1.62 1.97
22.53 0.78 2.20 1.61
23.98 1.20 1.77 1.13
18.96 0.83 1.17 1.77
24.64 1.01 1.76 2.04
19.52 1.34 1.81 1.63
19.97 0.55 2.25 1.93
18.86 1.06 1.98 1.63
22.79 0.90 1.54 2.27
20.38 1.88 1.15 1.45
16.51 0.51 1.19 1.93
18.32 0.53 1.75 1.44
16.04 1.21 1.01 1.46
18.91 0.90 1.68 1.66
20.09 0.97 1.72 1.59
19.83 1.01 1.72 1.42
18.42 1.29 1.47 2.05
21.99 1.35 1.94 2.28
20.57 1.07 2.31 1.77

So Basically i have calculated the regression equation to be Y-Hat = 12308.39 + 0.56987(Direct) + 3.32002(Newspaper) + 0.73218(Television)

Following this, there is a question regarding if direct = $800 Newspaper = $1200 and Television was equal to $2000 what would be the predicted gross sales for the store, i correctly calculated this to be equal too $18213...

Now the question i'm stuck on is the following:

Estimate the mean weekly gross sales for all stores that spend $800 on direct mailing, $1200 on newspaper advertising and $2000 on television commercials

What am i supposed to do here? i think it has something to do with sub-sampling the data set provided by the answer in the back of the book comes out as $18218, and i am getting an answer no where near to this...

PLEASE HELP!!!!

 

[utsstudent12]

Hey i'm doing the same assignment and i don't know how to answer that question either! But what did you get for (i) what statistics do you check for model performance?

 

[UTS101]

i have legit been sitting here trying to work out the answer to g... did they say anything in the powerpoints about regression performance?? i have been asleep in like all the lecs

 

[utsstudent12]

no i've never heard them say anything about performance! I don't know which statistics to use...

 

[UTS101]

i have no idea...

 

[Ethicks]

nvm

 

[utsstudent12]

i think we have to get a point estimate and + and - the margin of error! but im only guessing at this point!

 

[UTS101]

im not sure... look on uts online at the forum there the tutor talks about making a sub-sample or something... so i went in and singled out all the samples that spend over the amount in the question, and then calculated a new average but still got nowhere near the listed answer... its so frustrarting

 

[waterlml]

i think we have to get a point estimate and + and - the margin of error! but im only guessing at this point!

i dont get g too, have no idea how to do it. btw do we just use correlation coefficient for e??

 

[waterlml]

im not sure... look on uts online at the forum there the tutor talks about making a sub-sample or something... so i went in and singled out all the samples that spend over the amount in the question, and then calculated a new average but still got nowhere near the listed answer... its so frustrarting

i think the answer in our book is wrong...

 

[utsstudent12]

what are the statistics you check for model performance?

 

[InteGrand]

what are the statistics you check for model performance?

There is some useful information here (and probably many other websites): http://en.wikipedia.org/wiki/Regression_model_validation

 

[waterlml]

what are the statistics you check for model performance?

you can make the summary output in excel first

 

[UTS101]

Pretty sure r2 and adjusted r2 are measures of model performance, but what value is good

 

[InteGrand]

Pretty sure r2 and adjusted r2 are measures of model performance, but what value is good

An closest to 1 is best. However, looking at alone isn't enough to evaluate model performance, as that Wikipedia link shows. You should also analyse the residuals.

 

[waterlml]

An closest to 1 is best. However, looking at alone isn't enough to evaluate model performance, as that Wikipedia link shows. You should also analyse the residuals.

i only analysed r square 0.0......how do you analyse residuals, i'm not sure if we learnt anything about the residuals