Exponential Smoothing in Excel (Find α) | Use Solver to find smoothing constant alpha

[@kamalpatro5065]

One of the best videos available on KZbin! Shame it isn't reaching a lot of people in need of videos like these

[@Стрелок-ш7с]

This man explaining things to me that my 15k a year MBA teacher can't. Drop a link I'd love to donate.

[@joshemman]

Thanks for the love.

[@EllaMoss-i1l]

I never leave comments but i just had to say this video was so useful so thank you!

[@joshemman]

Thanks, Ella Moss.

[@braed23]

This is an absolutely fantastic video and it helped me immensely. Well narrated, clear and concise, THANK YOU SO MUCH!!

[@the_melon_2468]

Thanks Alot, this video really helped me out in school

[@CryptoFinanceIQ]

This video helped with my business analytics module thank you!

[@tanushreesingh4822]

your accent makes it so much better, you sound so cute and thank you for the amazing explanation

[@NoName-yr3nu]

OMG thank you so much. I've been struggling to find the best alpha using trial and error method ( as my teacher recommends) and this helps me a lot in my final BA exam next week.

[@sethcude25]

You are a saint

[@kk99807]

Extremely clear. Very helpful! THANK YOU!

[@ismailnassar4103]

Appreciated Your excellent explantion .

[@joshemman]

@@ismailnassar4103 Thanks Ismail 😊

[@michael654]

Thanks for these videos. I watched your whole forecasting playlist and found it very helpful!

[@Muhammed-MUH]

That was absolutely helpful 🙏

[@zahraash7002]

Thanks so much 🌷😊

[@leonidas4998]

Is it possible to get just a zero for alpha? Resulting in same number of Month1 demand for all the forecasts?

[@zying1925]

same question

[@mehgulanharikrishnan406]

Hi I have a question. So in conclusion if we were to choose only one best alpha value to do the exponential smoothing forecast, which alpha value of the 3 alpha value that we must use and why?

[@kaustavsengupta8757]

You should make Udemy course on excel prediction algorithm....this is gold

[@dheatrullyanisavitri8337]

why did you use GRG nonlinear for solving method, sir?

[@joshemman]

We use non-linear because the relationship demand is not either consistently increasing or decreasing as you go from month to month.

[@ANNYEONGx]

MSE, you need to take (Total Error ^ 2 / n -1 )

[@ezrasplace7504]

Hello, I hope you are still active.. I would like to know how to solve for alpha and beta for double exponential smoothing and I believe optimization also helps. Cant find any useful tutorials in youtube though

[@asrajo5728]

Thanks for sharing, usefull!

[@Day.One.Patch.]

The forecast produced (column D) is directly effected by what is put in cell D2 for 1-alpha. How do we decide what figure to place in here as you did not explain this aspect in the video?

[@joshemman]

Sorry, the formula for D2 is = 1 - C2.

[@jakeflane2218]

Great video on how to do it. Curious though, how would you get alpha in the first place?

[@78kurenai]

I have same question... it says on title find alpha....

[@kamalpatro5065]

It is mostly available on word, where you can go to insert and then special characters where you can find alpha

[@josephdaquila2479]

Alpha is just a smoothing coefficient you choose before creating each model. You would want to choose the alpha value that minimizes the error. This can be done through trial and error or any other optimization method (optimization because we're minimizing). Edit: Starting around here, he describes this process kzbin.info/www/bejne/eWatkISJh5yGipI&feature=shares&t=162 It depends on whether you choose MAD or MSE or some other form of error as the error you are wanting to minimize. I am not sure how you choose this.

[@eddydubois4584]

Question; when running solver I'm getting an alpha result of 0.39359 rather then the .30841 that you are getting. I have triple checked all the data is being input correctly the same as you have. The MAD MSE & MAPE scores are all off as well. :/ Not sure what's going on here.

[@joshemman]

You should obtain the same result. In cell D2, did you enter the formula =1-C2 ?

[@colisilepatriciadlamini8066]

Well presented, thank you

[@joshemman]

Glad it was helpful!

[@abrahamphilip6843]

is the demand column the exponential smoothing column

[@joshemman]

No. The Forecast column is the exponential smoothing column.

[@chisomfinbarrs5164]

So please how can this help me predict for upcoming months?

[@tansutazegul8297]

Sir, how can we determine what the alpha is?

[@tansutazegul8297]

I have figured it out. Through optimization👍

[@MrToxiclaughter]

@@tansutazegul8297 what?

[@KimTungs]

How can you find the best value for alpha?

[@shivamshankhdhar89]

Good video.

[@h.s.3883]

I wished our professor would take the time to show us how to do this rather than saying read the book and good luck.

[@elluranands6904]

Sir, while minimizing MAD, you opted for Nonlinear optimization while MAD is not non-linear. MSE is non-linear because it is raised to power. Kindly check whether we can use non-linear optimization for MAD?

[@joshemman]

Yes. They should be all non-linear.

[@elluranands6904]

@@joshemman Sorry Sir. That was wrong question. It should have been "Kindly check whether we can use linear optimization for MAD"

[@joshemman]

@@elluranands6904 You can, but the model is not linear. Thus, using linear optimization may not result in the best alpha.

[@elluranands6904]

@@joshemman Calculation of MAD is linear, right sir. While MSE is a quadratic function which is a non-linear. So, for MSE we can use non-linear optimization. How can we use non-linear optimization for MAD which is a linear function, Sir? Just curious to know.

[@joshemman]

@@elluranands6904MAD depends on on the patten in the data which may not be linear. The fact that it is not squared does not make it linear.

[@Sandycxh]

does it means that i can assume any value as long as it is more than 0 and less than 1 for Alpha?

[@joshemman]

Yes, you can start with any alpha value between 0 and 1.

[@Sandycxh]

@@joshemman omg thank you very much !! This video has been so useful !!

[@KimTungs]

So good but i wonder how to determine alpha at the beginning

[@alongtheway3840]

QUESTION: How did you conclude that 1 - alpha = 0.6? Is it always 0.6? For example, if we are starting with an alpha of 0.4, what would the 1- alpha be? Thank you for your helpful insight.

[@joshemman]

0.6 (1 - alpha) is obtained by a formula placed in cell D2. You can see the formula (1 - C2) from 0:00 to 0:40 in the address bar .

[@barkbent4324]

It depends on the alpha variable. The more you want to account for error, the greater the alpha. The greater the alpha, the lesser the (1-alpha). Fore example, If the optimal alpha is 0,1 then (1-a) is 0,9. [1 - 0,1 = 0,9]

[@hebaalawneh7998]

thank you so much , GREAT

[@zying1925]

why did i get 0 for alpha? Is it correct if I get 0 for alpha?

[@joshemman]

Sometimes you might need to show more decimal places. Otherwise, something could be wrong with setup.

[@srilaakha8865]

what if the demand for month 7 isn't given , how do I find MAD ?

[@joshemman]

You need both actual and forecast to calculate errors. If you don't have both, you cannot include that period in your MAD calculation.

[@srilaakha8865]

@@joshemman im suppose to calculate MAD for these 3 time series forecasting methods to figure out which is best however for my exponential smoothing - I only got demand and forecast for period 2-23 for my moving average - I only got from period 6-23 for my linear trend - I only got 1-23 therefore to calculate MAD do I calculate from period 6-23 for all forecast methods so I can compare equally ?

[@joshemman]

​@@srilaakha8865 Normally, you just calculate MAD based on all the periods you can find errors for. However, in some courses, while comparing methods, they limit the periods used to the periods in the method with the least number of errors. If that’s done in your course, then you should use 6 to 23.

[@barkbent4324]

Professional!

[@ilhankey7815]

thanks!

[@uwenolte8578]

Nice!

[@berend_simons]

i have no idea what any of this is about :D

[@High_Priest_Jonko]

Come on Berend, get it together

[@hassanbeila62]

Thaaaaank you man