ARIMA Using double seasonality
Handled using seasonal dummies and complete procedure is made using R. I have attached the description and my background.
Tableau de clarification publique
-
sunny2dutta
- il y a 9 ans
https://docs.google.com/spreadsheets/d/1gPDQOyRi8_x8ayYLlgw1Tvc2bVqvJyQTd6TrXoRDc3Y/edit#gid=0
- il y a 9 ans
-
Titulaire du concours - il y a 9 ans
you could post a link or something
- il y a 9 ans
-
Titulaire du concours - il y a 9 ans
By 01 - 03.2014 I meant January - March 2014 full month predictions
- il y a 9 ans
-
sunny2dutta
- il y a 9 ans
I cant comment with more than 1000 characters. 1st column is the month, 2nd col is hour 1-24, 3rd col day of week 7 for sat ,1 for sunday... Then we have the prediction
- il y a 9 ans
-
sunny2dutta
- il y a 9 ans
1 1 7 0.244462993
1 2 7 0.122951387
1 3 7 0.067111458
1 4 7 0.039741804
1 5 7 0.016439213
1 6 7 0.003933204
1 7 7 0.014462996
1 8 7 0.033915297
1 9 7 0.127596301
1 10 7 0.244721628
1 11 7 0.267231145
1 12 7 0.2844329
1 13 7 0.283295922
1 14 7 0.269164566
1 15 7 0.269393195
1 16 7 0.259231916
1 17 7 0.240079404
1 18 7 0.245854848
1 19 7 0.256316478
1 20 7 0.304676664
1 21 7 0.352555267
1 22 7 0.354398848
1 23 7 0.353268445
1 24 7 0.362814842
1 1 1 0.316394467
1 2 1 0.195316399
1 3 1 0.139449141
1 4 1 0.11208121
1 5 1 0.08877851
1 6 1 0.0684061
1 7 1 0.057876307
1 8 1 0.1062546
1 9 1 0.199935604
1 10 1 0.317060931
1 11 1 0.339570448
1 12 1 0.356772203
1 13 1 0.355635225
1 14 1 0.341503869
1 15 1 0.341732498
1 16 1 0.331571219
1 17 1 0.312418707
1 18 1 0.318194151
1 19 1 0.328655781
1 20 1 0.377015968
1 21 1 0.424894571
1 22 1 0.426738152
1 23 1 0.425607749
1 24 1 0.435154146- il y a 9 ans
-
Titulaire du concours - il y a 9 ans
can you provide a forecast for 01 - 03.2014, in a excel file in the same format as the provided consumption history, so we can compare with the actual. Please assume there are no hollidays in this period (just weekends)
- il y a 9 ans
-
sunny2dutta
- il y a 9 ans
1. For proper forecasting I would require to know which days are declared as holiday and which are not. I can go ahead and assume just Saturday and Sunday and make the prediction.
- il y a 9 ans
-
sunny2dutta
- il y a 9 ans
2. I worked on predicting sales for various e commerce websites with the help of google trends data. The sales had seasonality (double and single) effects of holiday seasons like chistmas and of course google trends data. The fund would invest money based on the predicted sales. 3. No the forecast is different for different hours in a day. If Monday and Wednesday are both working days( not holidays) the prediction is same for the same hour in the same month. If Monday is a holiday and Wednesday is not in a month the prediction is different. The effect of days of the week can easily be incorporated with another seasonality.
- il y a 9 ans
-
sunny2dutta
- il y a 9 ans
Hi Thanks for your feedback
- il y a 9 ans
-
Titulaire du concours - il y a 9 ans
hi, thx for your entry. I have a few questions: 1. Can you provide a forecast example based on your logic and on our data? 2. You said you worked for a hedge fund. What was your result there? 3. Your forecast assumes that the consumption is constant every hour in a certain day? or Does it assume that the consumption for example on Monday is the same as on Wednesday?
- il y a 9 ans
