Hello Mr. Piyush, Very nice video! I just have one question, though.. How will I calculate my seasonal indices when I do not have a complete set of data points? For example, if I am using monthly sales data for forecasting, and I have only 30 months of previous sales data, how will I calculate the seasonal indices? My data starts in Feb,2014 and ends in July,2017.
really thank you so much it helped me a lot for my MGT 300 final project, however i have a question that do we have to use a forecasting technique to get the predicted total demand for 2020?
A big H E L L O from COLOMBIA. I do something like you.....related to milking production, but using percentage....I know you are laughing now but, is my way...and it works with cow´s milk production. I will use your method. G R A C I A S M I L !!!!
Thank you for your video. Can we first deseasonalize the demand data and do the forecast based on any of the forecasting methods such as exponential smoothing etc and then apply the seasonality index?
Sorry for the late reply. I somehow missed your message. The 570 is a number I have assumed here. The yearly sale could have been obtained by any time series method. The yearly demand is not affected by seasons as all versions of seasons are inside the year and hence calculating that is easier.
he had made mistake..instead of 570 , 143.34 average of 1st quarter of all 4 years need to be taken and multiply it by seasonality factor of 1.0305 to get forecast of quarter 1 and so on for quarter 2, 3,4 .
Is it possible to use the demand figures for year 3 multiplied by the seasonal indices to predict year 4 instead of assuming 570 as you did? I actually did use the excel triple smoothing model and got slightly higher forecast figure than these values..Thanks!
Yes you may. But, it will get very complex like that. I would work at montly levels and then further disaggregate the monthly forecast into weeks. And, you could consider this app for monthly forecasting: zerohour.shinyapps.io/ETS_ARIMA_Prophet_monthly/
For monthly seasonal pattern, how do an analyst calculate the average price change? Do they total all the gains, and get the average from that? Or do they total all the losses....... I am not sure. What is it if u don't mind me asking? Thx.
@@ricardoguzman7518 You would need monthly or quarterly demand forecasts for production or fullfilment. it is a normal process to disaggregate forecasts like this from some aggregate forecast.
dear sir, thank u for the video..there is need of little correction in the tutorial.. instead of 570 , 143.34 average of 1st quarter of all 4 years need to be taken and multiply it by seasonality factor of 1.0305 to get forecast of quarter 1 and so on for quarter 2, 3,4 .
hello Rishu, thank you for this comment. The actual demand numbers already have the influence of seasons. So, averaging them and adding the effect of seasons again on them would not be valid. The ideal way is to use some method to find deseasonalized demand forecast and then add the effect of the season. I have used aggregate demand forecast from an unmentioned method as the deseasonalized demand. There can be many other ways to arrive at the deasonalized demand other than what I have done. Hope this helps.
I am not sure if you can use these methods to directly predict the demand for next Christmas. While you could use the Holt-Winter's model as a base, you would have to add contextual factors to account for current year.
You would have to use some alternate method with yearly data and find the yearly demand (=570). And, then 540/4 (-142.5) is the deseanalized average quarterly demand.
There are pdfs online that discuss Croston method. For example this: courses.edx.org/asset-v1:MITx+CTL.SC1x_1+2T2015+type@asset+block/w4l2_NewIntermittentProducts_ANNOTATED_FINAL.pdf. I have never used this method am not sure how exactly to use this.
@@piyushashah1 Thank you for responding, but can you please tell me why wont you trust its accuracy? You see I am new to forecasting, and I wanted to know if a company only has past data for 1 year, how should it go about forecasting for future months mitigating the seasonal effects?
@@smaranrai4898 There could have been some random variation in a specific period. If you just have one year data, you will assume that the random variation is a part of seasonality. However, if you have multi-period data, such random variations tend to average out and we get more accurate seasonality indices.
How to choose the correct forecast and do accuracy for retail? & How to do SKU level forecasting for the short term which includes seasonal items? I have sent mail. please check
![[LIVE] : ONE ลุมพินี 78 | คู่เอก "ปกรณ์ vs ฟาบิโอ"](http://i.ytimg.com/vi/LSOmgl0Th2c/mqdefault.jpg)
Watching your vids again! 10 years working in supply chain, but your videos always give a great reminder!
Thanks for writing this comment Bryce.
This was extremely helpful, and served to clear up my confusions ahead of my Operations Management final. Thank you!
Am happy that the video helped you. Thanks for leaving this comment.
Thank you for very nice video. I am a beginner and autodidactive learner. It is very helpful
Thank you for your wonderful and clear explanation!
Thanks for detail explanation. But may I know how we got the 4th year # as 570.(@5:00 min)
Thank you so much. It helped me analyse data for my PG Thesis.
Thanks so much. Very easily illustrated
Thank you!!! You explain things so well and make it easy.
+Andy Ryan Thanks for leaving the comment. All good things are always simple!!
Thank you sir
Thank you for your video!
Thank you! this was tremendously helpful :)
Do look at the other videos on forecasting as well.
nice
Hi Piyush, can please calculate the fitted values using the model to calculate variance analysis and also how did you get total of 570 for year 4?
Excellent !!!! Thank you so much
thank you. this video is very helpful
Hello Mr. Piyush,
Very nice video! I just have one question, though..
How will I calculate my seasonal indices when I do not have a complete set of data points? For example, if I am using monthly sales data for forecasting, and I have only 30 months of previous sales data, how will I calculate the seasonal indices? My data starts in Feb,2014 and ends in July,2017.
Thank you for your help! Was very informative.
really thank you so much it helped me a lot for my MGT 300 final project, however i have a question that do we have to use a forecasting technique to get the predicted total demand for 2020?
Very helpful. Thank you so much
A big H E L L O from COLOMBIA. I do something like you.....related to milking production, but using percentage....I know you are laughing now but, is my way...and it works with cow´s milk production. I will use your method. G R A C I A S M I L !!!!
Nice video, Sir please make videos using same example for weighted moving average and exponential smoothing.
Thank you for your video. Can we first deseasonalize the demand data and do the forecast based on any of the forecasting methods such as exponential smoothing etc and then apply the seasonality index?
where do you get the 570 from around the 5min mark?
Sorry for the late reply. I somehow missed your message. The 570 is a number I have assumed here. The yearly sale could have been obtained by any time series method. The yearly demand is not affected by seasons as all versions of seasons are inside the year and hence calculating that is easier.
he had made mistake..instead of 570 , 143.34 average of 1st quarter of all 4 years need to be taken and multiply it by seasonality factor of 1.0305 to get forecast of quarter 1 and so on for quarter 2, 3,4 .
Really helpful, thanks a lot!
Thanks for leaving this comment. Happy that the video was useful to you. Do also check the video on Measuring Forecasting Accuracy.
Is it possible to use the demand figures for year 3 multiplied by the seasonal indices to predict year 4 instead of assuming 570 as you did? I actually did use the excel triple smoothing model and got slightly higher forecast figure than these values..Thanks!
Thank you! Very helpful!
Hi jus wanna say thank you!
Hi, can we use the same method for calculating indices at weekly level throughout the year?
Yes you may. But, it will get very complex like that. I would work at montly levels and then further disaggregate the monthly forecast into weeks. And, you could consider this app for monthly forecasting: zerohour.shinyapps.io/ETS_ARIMA_Prophet_monthly/
For monthly seasonal pattern, how do an analyst calculate the average price change? Do they total all the gains, and get the average from that? Or do they total all the losses....... I am not sure. What is it if u don't mind me asking? Thx.
Do i have to predict year 4 or there is some calculation to get it ?
thanks!
Where that 570 for year 4th come from?
What would be the best method if the data does not show seasonality?
Thanks for sharing
Thanks for leaving the comment Aay...there are other forecasting videos as well, do have a look.
hello Mister Pitush .. where did you get the 570 demand for year 4?
+Rosario Dizon That is assumed to be given. Maybe some other method was used to derive it.
@@piyushashah1 Why would someone try to figure out the forecast if that amount was already given?
@@ricardoguzman7518 You would need monthly or quarterly demand forecasts for production or fullfilment. it is a normal process to disaggregate forecasts like this from some aggregate forecast.
dear sir,
thank u for the video..there is need of little correction in the tutorial..
instead of 570 , 143.34 average of 1st quarter of all 4 years need to be taken and multiply it by seasonality factor of 1.0305 to get forecast of quarter 1 and so on for quarter 2, 3,4 .
hello Rishu, thank you for this comment. The actual demand numbers already have the influence of seasons. So, averaging them and adding the effect of seasons again on them would not be valid. The ideal way is to use some method to find deseasonalized demand forecast and then add the effect of the season. I have used aggregate demand forecast from an unmentioned method as the deseasonalized demand. There can be many other ways to arrive at the deasonalized demand other than what I have done. Hope this helps.
i'm wondering how you can forecast daily volumes for christmas, using trends from past 3 years, as christmas day varies per year
I am not sure if you can use these methods to directly predict the demand for next Christmas. While you could use the Holt-Winter's model as a base, you would have to add contextual factors to account for current year.
Mr. Piyush can I ask what book or theory references you used for this method?
Hi piyush, how you get the value of 570?
That is an assumed number Anshul. Typically we could use some form of time series of regression to arrive at it. I have a video on that.
how the expected requirement demand 570 and ave 142.5 calculated
You would have to use some alternate method with yearly data and find the yearly demand (=570). And, then 540/4 (-142.5) is the deseanalized average quarterly demand.
Many thanks for this video!
Just dropped an email =)
Hi Piyush, can you also show us how to deseasonalize the seasonality trends.
Thanks
HI Swati, I'll try. Thanks a lot for the suggestion.
How did you find the total for the fourth year (570)
This video was so good up until the 5 min mark. Year 4 being introduced with no explanation for the 570 demand......
It is an assumed number that we could have obtained from regression or any other method.
Sir how do you model the croston method for intermittent demands?
There are pdfs online that discuss Croston method. For example this: courses.edx.org/asset-v1:MITx+CTL.SC1x_1+2T2015+type@asset+block/w4l2_NewIntermittentProducts_ANNOTATED_FINAL.pdf. I have never used this method am not sure how exactly to use this.
How would calculate seasonal indices using average percentage method?
Do you mean percentage of sales method? Why do you need to do that? Can you give me details please?
Hi! can we find the seasonal indices for data that only have 1-year of data?
You can... But I won't trust the accuracy of such one year based indices.
@@piyushashah1 Thank you for responding, but can you please tell me why wont you trust its accuracy? You see I am new to forecasting, and I wanted to know if a company only has past data for 1 year, how should it go about forecasting for future months mitigating the seasonal effects?
@@smaranrai4898 There could have been some random variation in a specific period. If you just have one year data, you will assume that the random variation is a part of seasonality. However, if you have multi-period data, such random variations tend to average out and we get more accurate seasonality indices.
@@piyushashah1 Thank you so muchh piyush!! This makes things much clearer!!!
570 year 4 it is immaginary?
As mentioned in the video, you can use some other method to forecast that.
Why not test the data from forecast eror ? How we know that forecast is well ?
Yanuar Anaba Wahyuesa We should check the Forecast for errrors. There is a separate video on the different ways to calculate the errors.
what if I donot have 4th year total data 570. How to forecast then?
+ankushzap With this method you need that '570'. We can use some other method, maybe regression to get that.
How this 570 came?
I have assumed that number for this video. We can get it from moving average, regression or any of the other methods explained in other videos.
how do we get the 570
That is an assumption for this video. In actual scenario we can get it from linear regression or some mechanics performed on yearly demand data.
what is solution? when Anova result is not significant .
Asima Shahzadi sorry, can you elaborate on your question?
Great click noise
Daniel Burke Yes, and sorry. But the free tool I have has this limitation. I hope I can get a better method or a tool.
How to choose the correct forecast and do accuracy for retail? & How to do SKU level forecasting for the short term which includes seasonal items? I have sent mail. please check