Readings for this week...

[Gabriel Bowers]

 

Aside from the "Diffusion Life Cycle Model" slides, do we have any other required reading for this week?  I thought Heineke had mentioned that we should read certain chapters, however, he has not indicated any chapters in the syllabus.

Gabriel

[Spencer]

It's on the bottom of that page. Ch 5 and 6

Sent from my iPhone

[Spencer Hsu]

by the way, is the case inside the book or a pdf? what page is it in the book if it's in the book.

[alex smirnov]

is there case assigned?   G-O-D Heineke is confusing  ;)

[jonathan car]

My answers to Problem 9 (Sales Forecasting) are per attached PDF. Please check and let me know if you agree or disagree, and tell me why. THanks

Regards 

Jonathan

---

From: alex smirnov <alex.s...@gmail.com>
To: econ_401_w...@googlegroups.com
Sent: Sun, February 13, 2011 9:22:17 PM
Subject: Re: Readings for this week...

[Spencer]

When is this due? I thought it was due 2 weeks from today. That's what it said on the syllabus

Sent from my iPhone

<Problem 9 Forecasting Sales (Thompson).pdf>

[alex smirnov]

It's due on Wed, and it's PROBLEM 8 (along with cable car fares problem), not problem 9, or am I going crazy?

[Spencer Hsu]

I think it's Problem 9 Homework 8. Heneike's Syllabus is off here. No wonder it was confusing.

[Caroline Poncet Chapman]

Quick question: in the Excel file provided on Eres for the logistic problem, does the Sales column refer strictly to sales in that time period, or is it already cumulative - St? Wish I had asked him in class...

Thank you in advance to anyone who can answer :)

[Spencer Hsu]

Hey Jonathan, how did y ou eliminate the intercept?

[Spencer Hsu]

Nevermind got it. 

[Elliott Le]

What did you guys try? trendline or just plugging in one of the data points?

Elliott Le

--
Elliott Le

[Gabriel Bowers]

Trendline for a+b*t. I built one from four quarters of data for q1 and
q2. Then used five quarters of data for q3.

>>>>>> ------------------------------
>>>>>> *From:* alex smirnov <alex.s...@gmail.com>
>>>>>> *To:* <econ_401_w...@googlegroups.com>
>>>>>> econ_401_w...@googlegroups.com
>>>>>> *Sent:* Sun, February 13, 2011 9:22:17 PM
>>>>>> *Subject:* Re: Readings for this week...

>>>>>>
>>>>>> is there case assigned? G-O-D Heineke is confusing ;)
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Sun, Feb 13, 2011 at 6:21 PM, Spencer Hsu < <sph...@gmail.com>
>>>>>> sph...@gmail.com> wrote:
>>>>>>
>>>>>>> by the way, is the case inside the book or a pdf? what page is it in
>>>>>>> the book if it's in the book.
>>>>>>>
>>>>>>>
>>>>>>> On Sun, Feb 13, 2011 at 5:20 PM, Spencer < <sph...@gmail.com>
>>>>>>> sph...@gmail.com> wrote:
>>>>>>>
>>>>>>>> It's on the bottom of that page. Ch 5 and 6
>>>>>>>>
>>>>>>>> Sent from my iPhone
>>>>>>>>
>>>>>>>> On Feb 13, 2011, at 5:14 PM, Gabriel Bowers
>>>>>>>> <<gabrie...@gmail.com>
>>>>>>>> gabrie...@gmail.com> wrote:
>>>>>>>>
>>>>>>>> >
>>>>>>>> > Aside from the "Diffusion Life Cycle Model" slides, do we have any
>>>>>>>> other required reading for this week? I thought Heineke had
>>>>>>>> mentioned that
>>>>>>>> we should read certain chapters, however, he has not indicated any
>>>>>>>> chapters
>>>>>>>> in the syllabus.
>>>>>>>> >
>>>>>>>> > Gabriel
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> <Problem 9 Forecasting Sales (Thompson).pdf>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
>
> --
> Elliott Le
>

--
Sent from my mobile device

[Carla Nunes]

Hi guys,

 

Quick question on how to tackle this... we basically run a regression on the column that will have the calculated ln (St / m-St).  This column will be our LRS variable, and t will be our RHS variable.  Correct?

 

Do we need to run a different regression besides that one?

 

Thx,

Carla

[Gabriel Bowers]

ln(s/m-s) shows a linear relationship with t. Therefore, you can fit
'a+bt' from this relationship.

[Carla Nunes]

Did you guys get this:

 

Intercept	-8.263946887
t	0.825933324

so, ln (S/(m-S) = a + bt = -8.26 + 0.826 t

 

Please let me know!  Thx,

 

Carla

[Spencer]

So is jonathan's then?

Sent from my iPhone

[Carla Nunes]

No, I didn't get the same as Jonathan.  In fact, I'm not sure why he decided to remove the intercept, and why the two regressions.

[Spencer Hsu]

It seems he removed it based on the very high p values.

[alex smirnov]

Carla, my numbers are the same as yours

I wouldn't read too much into Jonathan's answers as they were done before yesterday's lecture

btw, I hope y'all don't forget to get the critical values for both F and T (excel functions) , as well as do the 3 residual tests (Durban Watson, normalcy, and one other one I can't remember)

[Carla Nunes]

Ok, good, Alex.  At least we got same numbers for the regression.  Once I'm done w/ the other calculations, I'll post them here to check again.

[Carla Nunes]

Guys, here are my answers for Thompson:

 

a) regression model w/ coefficients

 

Intercept	-8.263946887
t	0.825933324

b) sales in 5th quarter = 1513.13;  %error = 3.145%

c) based on graph, model predicts well

d) model's prediction = 83.85%, so %error = 11.8%

 

Let me know if you guys got the same.  

 

See ya,

Carla

[jonathan car]

Got the same answers as Carla. Refer to attachment.

 

Regards

Jonathan

---

From: Carla Nunes <ccrn...@gmail.com>
To: econ_401_w...@googlegroups.com
Sent: Tue, February 15, 2011 12:55:53 PM
Subject: Re: Readings for this week...

[Adi Aloni]

Hi,

Notice that 1513 is the cumulative sales up to the 5^th quarter. You have to subtract the cumulative sales up to the 4^th quarter to get the numbers for the 5^th quarter alone.

I think we’re supposed to calculate the model error according to the 5^th quarter alone.

Adi.

[Carla Nunes]

Good point, Adi.  I did the calculation for the cumulative values, not the quarter itself.  Let me update my sheet.

[Adi Aloni]

How did you calculate % error for part d? we don’t have actual numbers…

Is it compared to the 75% prediction?

[Carla Nunes]

Yes, that's what I did.  Not sure if it's correct though, so I sent an email to the TA to confirm.  Still waiting for her response.

 

For part a, I now got:  844.72.  error =  1.043%

Let me know if this is what you got.

 

Carla

[Carla Nunes]

Jonathan,

 

How did you get to 8.9% on your letter d?  I didn't get that.... mine is 11.8%.  I did (75% - 83.5%) / 75%

 

Let me know.

 

Carla

[Adi Aloni]

OK… here’s where it gets sticky…

What number do we use for calculating the forecasted stand alone Q5 sales? If we use the actual cumulative sales data up to the 4^th qtr, I got a forecast of 882 with 5.5% error. However, I’m not sure we should use the actual number, because in part c we calculate all 5 quarters based on the forecast.

What do you think? Is it worth emailing the professor/TA?

[Carla Nunes]

s(5) = S(5) – S(4), so for both S(5) and S(4), i use the model.  That will give us the sales for the quarter alone according to the model. 

 

In my case, I get S(5) = 1513.13, S(4) = 668.41, which gives me s(5) = 844.72

 

With that, you compare w/ the given sales for quarter 5, which is 836.

 

That's what I think it should be done, but again, never 100% sure. 

[jonathan car]

Carla is once again correct. My sF is calculated wrong. It should be
s(5) = S(5) – S(4),     instead of s(5) - S(5) - s(4)

 

Thank you Carla

 

Jonathan

---

From: Carla Nunes <ccrn...@gmail.com>
To: econ_401_w...@googlegroups.com

Sent: Tue, February 15, 2011 2:07:47 PM

[Adi Aloni]

Alex – are you sure we have to go through all the linear regression tests in this model? There are so few observations, it doesn’t make sense…

[alex smirnov]

i'm not sure........i just remember that Heineke mentioned that the TA went easy on the grading, but in the future he wants that stuff

perhaps the residual tests don't make sense.......but T-critical and F-critical values are still valid

[Carla Nunes]

Adi, Alex, which other regressions you're talking about?

[Carla Nunes]

This is what I got for the cable ride question:

 

Elasticity = -0.887 (inelastic), so they should raise prices.

 

Let me know if you have the same (or not)  :)

[Adi Aloni]

Hi guys,

I’m posting my question and the professor’s answer to it. Now tell me what tests I should include in my answers….

 

Q: In the logistic model, since we have so few observations, do we still calculate DW and check residuals and so on?
      In a broader context, is there any case in which we reject the linear regression results in this model?

 

A: Thanks Adi. I forgot to mention in class that one expects highly correlated residuals in a model like this. The only question is does it fit. If it fits poorly the model will do poorly

[Carla Nunes]

I don't know if I understand what he means.  I did the DW test and mine shows as 2.38, not an ideal number but, it is still within the 1.5 and 2.5 range, so I said it's not serially correlated.

[alex smirnov]

I got the same for DW.  and also said that it is within range, but hard to tell with 4 observations

how did you guys calculate the Elasticity for the cable car problem?

this is confusing.  One part talks about a 67% increase (to $5) which means that fares went from $3 to $5

Another part says that "cable car fares more than doubled"

which is it 67% or doubled?

And I can't figure out how I get the info on the change in demand

[alex smirnov]

alright, i got the same thing Carla got.

P0 = $5/1.67 = $3

Then simply express both Q1 and Q0 in terms of either R1 or R0 (R1 = 1.27R0) and put that into the Ep equation

Ep = ((Q1-Q0)/(P1-P0)*(P1/Q1))

now on to OMIS 355 homework...

[Robert Binkley]

For the Cable Car problem, calculating the Elasticity can also be done this way:

 

%∆Q = Ep * %∆P  where %∆Q = %∆R - %∆P

 

This gives the answer Ep = -.657 

I am not sure why this Ep is different than yours.

 

The hard part I am struggling with is how to pick the optimal price, we know it should be higher than 5 dollars.  A cost figure is provided so how do we maximize profit. 

Can we assume cost is fixed so all we need to do is set the price to maximize revenue at Ep=-1?  Do we need to create a demand function?

 

-rob

This email and any attachments are intended for the sole use of the named recipient(s) and contain(s) confidential information that may be proprietary, privileged or copyrighted under applicable law. If you are not the intended recipient, do not read, copy, or forward this email message or any attachments. Delete this email message and any attachments immediately.

[alex smirnov]

the problem doesn't ask to pick the optimal price.  I just say that they should further increase from $5 from a purely economic view

btw, where does %∆Q = %∆R - %∆P come from?  I'm not sure I get it conceptually

if your revenue go up by 5% and your price goes up by 3%, does that mean you sell 2% more of your product??

[Carla Nunes]

Hi Robert,

 

I used the formula:

 

∆Q/Q = Ep * ∆P/P, where ∆Q = Q1 - Q0, and ∆P = P1 - P0.

 

Since we don't have Q1 or Q0, we write those as a function of revenue (R0), which will get cancelled out in the Ep formula.

 

If you want, I can send you my calculations.

 

Also, I don't think we need to come up w/ a specific price for this answer.  We only need to say whether they should raise the price, lower it, or keep the same, and why.

 

Carla

On Wed, Feb 16, 2011 at 10:07 AM, Robert Binkley <Robert....@xilinx.com> wrote:

[Gabriel Bowers]

My answer for #2 is slightly different than Carla and Alex's Ep.  So I'll post my calculations:

 

- R2 = 1.23 R1

- P2Q2 = P1Q1*(1.23)

- $5 * Q2 = $2.99 * Q1 * 1.23

Therefore: Q2 = 0.736*Q1 or you could saw that (Q2-Q1)/Q1 = -26.5%

So given that P2 went up 67% over P1, the elasticity equation would be:

Ep = change Q / change P = -0.265/0.67 = -0.39.

 

Let me know if anyone can see a mistake in this calculation.

 

My other answer for the question relating to chapter 6 is relating to the cost function.  Given the higher costs for operating the cable cars that reducing Q by 26.5% should improve the overall profit for the cable car operation.

 

Gabriel

[Gayathri Ravi]

I got similar to Gabe's

[Kevin]

I'm not clear on how to write Q1 and Q0 as a function of revenue...
Can anyone clear this up?



Kevin

On Feb 16, 11:13 am, Carla Nunes <ccrnu...@gmail.com> wrote:
> Hi Robert,
>
> I used the formula:
>
> ∆Q/Q = Ep * ∆P/P, where ∆Q = Q1 - Q0, and ∆P = P1 - P0.
>
> Since we don't have Q1 or Q0, we write those as a function of revenue (R0),
> which will get cancelled out in the Ep formula.
>
> If you want, I can send you my calculations.
>
> Also, I don't think we need to come up w/ a specific price for this answer.
> We only need to say whether they should raise the price, lower it, or keep
> the same, and why.
>
> Carla
> On Wed, Feb 16, 2011 at 10:07 AM, Robert Binkley

> <Robert.Bink...@xilinx.com>wrote:

>
>
>
>
>
>
>
> >  For the Cable Car problem, calculating the Elasticity can also be done
> > this way:
>
> > %∆Q = Ep * %∆P  where %∆Q = %∆R - %∆P
>
> > This gives the answer Ep = -.657
>
> > I am not sure why this Ep is different than yours.
>
> > The hard part I am struggling with is how to pick the optimal price, we
> > know it should be higher than 5 dollars.  A cost figure is provided so how
> > do we maximize profit.
>
> > Can we assume cost is fixed so all we need to do is set the price to
> > maximize revenue at Ep=-1?  Do we need to create a demand function?
>
> > -rob
>

> > *From:* econ_401_w...@googlegroups.com [mailto:
> > econ_401_w...@googlegroups.com] *On Behalf Of *alex smirnov
> > *Sent:* Tuesday, February 15, 2011 9:08 PM
>
> > *To:* econ_401_w...@googlegroups.com

> > *Subject:* Re: Readings for this week...
>

> > alright, i got the same thing Carla got.
>
> > P0 = $5/1.67 = $3
>
> > Then simply express both Q1 and Q0 in terms of either R1 or R0 (R1 =
> > 1.27R0) and put that into the Ep equation
>
> > Ep = ((Q1-Q0)/(P1-P0)*(P1/Q1))
>
> > now on to OMIS 355 homework...
>

> > On Tue, Feb 15, 2011 at 6:16 PM, alex smirnov <alex.smir...@gmail.com>

> > wrote:
>
> > I got the same for DW.  and also said that it is within range, but hard to
> > tell with 4 observations
>
> > how did you guys calculate the Elasticity for the cable car problem?
>
> > this is confusing.  One part talks about a 67% increase (to $5) which means
> > that fares went from $3 to $5
>
> > Another part says that "cable car fares more than doubled"
>
> > which is it 67% or doubled?
>
> > And I can't figure out how I get the info on the change in demand
>

> > On Tue, Feb 15, 2011 at 5:15 PM, Carla Nunes <ccrnu...@gmail.com> wrote:
>
> > I don't know if I understand what he means.  I did the DW test and mine
> > shows as 2.38, not an ideal number but, it is still within the 1.5 and 2.5
> > range, so I said it's not serially correlated.
>

> > On Tue, Feb 15, 2011 at 5:10 PM, Adi Aloni <aloni....@gmail.com> wrote:
>
> > Hi guys,
>
> > I’m posting my question and the professor’s answer to it. Now tell me what
> > tests I should include in my answers….
>
> > Q: In the logistic model, since we have so few observations, do we still
> > calculate DW and check residuals and so on?
> >       In a broader context, is there any case in which we reject the linear
> > regression results in this model?
>
> > A: Thanks Adi. I forgot to mention in class that one expects highly
> > correlated residuals in a model like this. The only question is does it fit.
> > If it fits poorly the model will do poorly
>
> > Adi.
>

> > *From:* econ_401_w...@googlegroups.com [mailto:
> > econ_401_w...@googlegroups.com] *On Behalf Of *Carla Nunes
>
> > *Sent:* Tuesday, February 15, 2011 3:38 PM
> > *To:* econ_401_w...@googlegroups.com

> > *Subject:* Re: Readings for this week...
>

> > This is what I got for the cable ride question:
>
> > Elasticity = -0.887 (inelastic), so they should raise prices.
>
> > Let me know if you have the same (or not)  :)
>

> > On Tue, Feb 15, 2011 at 3:14 PM, Carla Nunes <ccrnu...@gmail.com> wrote:
>
> > Adi, Alex, which other regressions you're talking about?
>

> > On Tue, Feb 15, 2011 at 3:04 PM, alex smirnov <alex.smir...@gmail.com>

> > wrote:
>
> > i'm not sure........i just remember that Heineke mentioned that the TA went
> > easy on the grading, but in the future he wants that stuff
>
> > perhaps the residual tests don't make sense.......but T-critical and
> > F-critical values are still valid
>

> > On Tue, Feb 15, 2011 at 2:46 PM, Adi Aloni <aloni....@gmail.com> wrote:
>
> > Alex – are you sure we have to go through all the linear regression tests
> > in this model? There are so few observations, it doesn’t make sense…
>

> > *From:* econ_401_w...@googlegroups.com [mailto:
> > econ_401_w...@googlegroups.com] *On Behalf Of *jonathan car
> > *Sent:* Tuesday, February 15, 2011 2:38 PM
>
> > *To:* econ_401_w...@googlegroups.com

> > *Subject:* Re: Readings for this week...
>

> > Carla is once again correct. My sF is calculated wrong. It should be
> > s(5) = S(5) – S(4),     instead of s(5) - S(5) - s(4)
>
> > Thank you Carla
>
> > Jonathan

> >  ------------------------------
>
> > *From:* Carla Nunes <ccrnu...@gmail.com>
> > *To:* econ_401_w...@googlegroups.com
> > *Sent:* Tue, February 15, 2011 2:07:47 PM

> > *Subject:* Re: Readings for this week...
>

> > s(5) = S(5) – S(4), so for both S(5) and S(4), i use the model.  That will
> > give us the sales for the quarter alone according to the model.
>
> > In my case, I get S(5) = 1513.13, S(4) = 668.41, which gives me s(5) =
> > 844.72
>
> > With that, you compare w/ the given sales for quarter 5, which is 836.
>
> > That's what I think it should be done, but again, never 100% sure.
>

> > On Tue, Feb 15, 2011 at 2:00 PM, Adi Aloni <aloni....@gmail.com> wrote:
>
> > OK… here’s where it gets sticky…
>
> > What number do we use for calculating the forecasted stand alone Q5 sales?
> > If we use the actual cumulative sales data up to the 4th qtr, I got a
> > forecast of 882 with 5.5% error. However, I’m not sure we should use the
> > actual number, because in part c we calculate all 5 quarters based on the
> > forecast.
>
> > What do you think? Is it worth emailing the professor/TA?
>

> > *From:* econ_401_w...@googlegroups.com [mailto:
> > econ_401_w...@googlegroups.com] *On Behalf Of *Carla Nunes
> > *Sent:* Tuesday, February 15, 2011 1:56 PM
>
> > *To:* econ_401_w...@googlegroups.com

> > *Subject:* Re: Readings for this week...
>

> > Yes, that's what I did.  Not sure if it's correct though, so I sent an
> > email to the TA to confirm.  Still waiting for her response.
>
> > For part a, I now got:  844.72.  error =  1.043%
>
> > Let me know if this is what you got.
>
> > Carla
>

> > On Tue, Feb 15, 2011 at 1:50 PM, Adi Aloni <aloni....@gmail.com> wrote:
>
> > How did you calculate % error for part d? we don’t have actual numbers…
>
> > Is it compared to the 75% prediction?
>

> > *From:* econ_401_w...@googlegroups.com [mailto:
> > econ_401_w...@googlegroups.com] *On Behalf Of *Carla Nunes
> > *Sent:* Tuesday, February 15, 2011 1:45 PM
>
> > *To:* econ_401_w...@googlegroups.com

> > *Subject:* Re: Readings for this week...
>

> > Good point, Adi.  I did the calculation for the cumulative values, not the
> > quarter itself.  Let me update my sheet.
>

> > On Tue, Feb 15, 2011 at 1:39 PM, Adi Aloni <aloni....@gmail.com> wrote:
>
> > Hi,
>
> > Notice that 1513 is the cumulative sales up to the 5th quarter. You have
> > to subtract the cumulative sales up to the 4th quarter to get the numbers
> > for the 5th quarter alone.
>
> > I think we’re supposed to calculate the model error according to the 5thquarter alone.
>
> > Adi.
>

> > *From:* econ_401_w...@googlegroups.com [mailto:
> > econ_401_w...@googlegroups.com] *On Behalf Of *Carla Nunes
> > *Sent:* Tuesday, February 15, 2011 12:56 PM
>
> > *To:* econ_401_w...@googlegroups.com

>
> > *Subject:* Re: Readings for this week...
>

> > Guys, here are my answers for Thompson:
>
> > a) regression model w/ coefficients
>
> > Intercept
>
> > -8.263946887
>
> > t
>
> > 0.825933324
>
> > b) sales in 5th quarter = 1513.13;  %error = 3.145%
>
> > c) based on graph, model predicts well
>
> > d) model's prediction = 83.85%, so %error = 11.8%
>
> > Let me know if you guys got the same.
>
> > See ya,
>
> > Carla
>

> > On Tue, Feb 15, 2011 at 11:16 AM, Carla Nunes <ccrnu...@gmail.com> wrote:
>
> > Ok, good, Alex.  At least we got same numbers for the regression.  Once I'm
> > done w/ the other calculations, I'll post them here to check again.
>

> > On Tue, Feb 15, 2011 at 11:10 AM, alex smirnov <alex.smir...@gmail.com>

> > wrote:
>
> > Carla, my numbers are the same as yours
>
> > I wouldn't read too much into Jonathan's answers as they were done before
> > yesterday's lecture
>
> > btw, I hope y'all don't forget to get the critical values for both F and T
> > (excel functions) , as well as do the 3 residual tests (Durban Watson,
> > normalcy, and one other one I can't remember)
>

> > On Tue, Feb 15, 2011 at 10:31 AM, Spencer Hsu <sphs...@gmail.com> wrote:
>
> > It seems he removed it based on the very high p values.
>

> > On Tue, Feb 15, 2011 at 10:21 AM, Carla Nunes <ccrnu...@gmail.com> wrote:
>
> > No, I didn't get the same as Jonathan.  In fact, I'm not sure why he
> > decided to remove the intercept, and why the two regressions.
>

> > On Tue, Feb 15, 2011 at 10:19 AM, Spencer <sphs...@gmail.com> wrote:
>
> > So is jonathan's then?
>
> > Sent from my iPhone
>

> > On Feb 15, 2011, at 10:08 AM, Carla Nunes <ccrnu...@gmail.com> wrote:
>
> >  Did you guys get this:
>
> > Intercept
>
> > -8.263946887
>
> > t
>
> > 0.825933324
>
> > so, ln (S/(m-S) = a + bt = -8.26 + 0.826 t
>
> > Please let me know!  Thx,
>
> > Carla
>

> > On Tue, Feb 15, 2011 at 9:56 AM, Gabriel Bowers <gabrielbow...@gmail.com>

> > wrote:
>
> > ln(s/m-s) shows a linear relationship with t.  Therefore, you can fit
> > 'a+bt' from this relationship.
>

> > On 2/15/11, Carla Nunes <ccrnu...@gmail.com> wrote:
> > > Hi guys,
>
> > > Quick question on how to tackle this... we basically run a regression on
> > the
> > > column that will have the calculated ln (St / m-St).  This column will be
> > > our LRS variable, and t will be our RHS variable.  Correct?
>
> > > Do we need to run a different regression
>

> ...
>
> read more »

[Carla Nunes]

My answer to the calculations is attached.  I'm not sure if it's correct, but that's what I'll be turning in.  :)

[alex smirnov]

Kevin,

R1 = Q1*P1  =>  Q1 = R1/Q1

R2 = Q2*P2  =>  Q2 = R2/Q2

R2 = 1.27R1

P1 = 5/1.67 = 3

P2 = 5

On Wed, Feb 16, 2011 at 12:50 PM, Kevin <jo.k...@gmail.com> wrote:

[Kevin]

Thanks. Can any point me in the right direction for the logistics
model? I seem to have gotten the regression part fine, but I'm stuck
after that.



Kevin

> ...
>
> read more »

[David Song]

kevin,

make sure your dependent variable (the thing you're trying to
forecast) is Ln{S(t) / [m - S(t)]}, then do the Algebra gymnastic
Heineke showed us in class to get "e" to the power of your equation,
e.g. e^-8.264+.826t, then inverse it to solve for forecasted
cumulative sales for that Quarter. see Jonathan's pdf for reference.

hope this helps
david

[alex smirnov]

Kevin, i'm not sure what you're asking.  You can call me at 415.412.6064.

[Carla Nunes]

Kevin,

 

Hope this helps:

 

1) Once you have found the coefficients a, b from the regression, you substitute them in the equation for the cumulative sales:  S(t) = m / 1 + e^ -(a+bt).  This becomes your forecasting model.

2) So, to calculate the sales level for quarter 5, you calculate forecasted S(5) and S(4) and subtract one from the other.

 

3) The percentage error for quarter 5 is calculated by:  [sales level actual - sales level forecasted (from part 2)] divided by sales level actual

 

4) For plotting F actual vs F forecasted, you will need to first calculate the value of F actual and F forecasted for all 5 quarters.  The formula for F forecasted is F(t) = 1 / 1 + e^ -(a+bt).  The formula for F actual is the actual cumulative level of sales S(t) divided by m (market size)

 

5) For the last question on Thompson, you'll need to calculate the F(12), this is the percentage of market penetration.  Then you have to calculate the % error the same way you did on part 3, but you don't have the actual, only the estimate of 75% by the CMO, so your formula would become  [75 - F(12) forecasted ] divided by 75.  Remember to take the absolute value.

[Kevin]

Thanks for the help guys. Carla your explanation really helped.


Kevin

On Feb 16, 2:22 pm, Carla Nunes <ccrnu...@gmail.com> wrote:
> Kevin,
>
> Hope this helps:
>
> 1) Once you have found the coefficients a, b from the regression, you
> substitute them in the equation for the cumulative sales:  S(t) = m / 1 + e^
> -(a+bt).  This becomes your forecasting model.
> 2) So, to calculate the sales level for quarter 5, you calculate forecasted
> S(5) and S(4) and subtract one from the other.
>
> 3) The percentage error for quarter 5 is calculated by:  [sales level actual
> - sales level forecasted (from part 2)] divided by sales level actual
>
> 4) For plotting F actual vs F forecasted, you will need to first calculate
> the value of F actual and F forecasted for all 5 quarters.  The formula for
> F forecasted is F(t) = 1 / 1 + e^ -(a+bt).  The formula for F actual is the
> actual cumulative level of sales S(t) divided by m (market size)
>
> 5) For the last question on Thompson, you'll need to calculate the F(12),
> this is the percentage of market penetration.  Then you have to calculate
> the % error the same way you did on part 3, but you don't have the actual,
> only the estimate of 75% by the CMO, so your formula would become  [75 -
> F(12) forecasted ] divided by 75.  Remember to take the absolute value.
>

> On Wed, Feb 16, 2011 at 1:43 PM, alex smirnov <alex.smir...@gmail.com>wrote:
>
>
>
>
>
>
>
> > Kevin, i'm not sure what you're asking.  You can call me at 415.412.6064.
>

[Spencer Hsu]

I learn like 3x more from these discussions than in class. Sad