Freelancer : vinaygupta88
Partager :
Signaler la proposition

Forecasting Analysis

Hello team I have significant expertise in forecasting as I have implemented excel based forecast model for at least 10 clients. Forecasting is a 7 step process 1. Removing outlier from data 2. Identify seasonal effects 3. Removing seasonality and trend 4. Forecasting using exponential smoothing 5. Adding back seasonality and trend 6. Distributing daily forecast into hourly forecast 7. Producing hourly forecast I have forcasted for 3 months in 2014. I am not able to share that file? Should I mail it to you? Forecast will improve once we have holiday information, external variables and missing data (May-Jun 2011). We would need to freeze on significant external parameter using regression analysis. Advantage: Easy to implement Good accuracy Disadvantage Equal weight age to all historical data Model will be built into excel which can be integrated to the C# plateform. I will also use VBA programming to automate all the steps I am ready to have a detailed call if that helps


                                                                                                                        Proposition n°                                            2
                                         du concours                                             Forecast hourly energy consumption for a certain period of time
Proposition n°2

Tableau de clarification publique

  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    thank you. I have reviewed it.

    • il y a 7 ans
    1. vinaygupta88
      vinaygupta88
      • il y a 7 ans

      Thanks, let us know next steps. we could also collaborate with other participants if needed. We can also fine tune our forecast if given more data around external variables, holiday list etc. Thanks

      • il y a 7 ans
  • vinaygupta88
    vinaygupta88
    • il y a 7 ans

    I have mailed the latest file which include forecast for Feb'14 and Mar'14.

    • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    can you provide forecast for 02 - 03.2014 as well so we can test further?

    • il y a 7 ans
    1. vinaygupta88
      vinaygupta88
      • il y a 7 ans

      Sure. we will be sending you the same by this Monday or Tuesday. Thanks

      • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    Ok, sorry for the last message, we tested with our current logic and you performed 50% better then our current logic. You are still in the game

    • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    Thank you for your effort though

    • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    it is way to high error

    • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    hi, thx for your forecast. Your error is Average error 79.09917698 %
    Min error 0 %
    Max error 454.3209877 %

    • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    please send your forecast to mihai.darzan ( a t ) ringhel.ro

    • il y a 7 ans
  • vinaygupta88
    vinaygupta88
    • il y a 7 ans

    Hi, I have analyized the data in excel. I have used data till Oct'13 to forecast Nov'13, Dec,13 and Jan'14. I am getting forcast accuracy of ~80% at daily level and 70% at hourly level for the month of Nov and Dec. As this is first cut analysis, we can expect the accuracy to jump by 5-10% when we get holiday information and data of external variable. I am not able to share excel file due to freelancer limitations. Let me know any email id where I can share the files? You can also compare Jan'14 forecast with original data

    • il y a 7 ans
  • vinaygupta88
    vinaygupta88
    • il y a 7 ans

    I would integrate the excel with SQL database (using ODB data connection) to get the data cuts as we require and then provide a forecast which client can view in excel itself. I will do some cosmetic changes to excel file to give a finished product look like a dashboard. Excel will be completely automated and with a press of a button client should be able to view forecast.

    If client want the output on the GUI built on C# platform only then only we would need to either integrating the excel with C# or write a separate code in C# without using excel. I think best is to align with client the results first using excel and then we can use that algorithm to build in C#

    • il y a 7 ans
    1. vinaygupta88
      vinaygupta88
      • il y a 7 ans

      Hi, The approach used for forecasting is not fixed. The basic method of forecasting is improvised according to the case. It is an art which requires some iterations and tweaking until we get the numbers that makes the client happy. Excel provides us the speed and flexibility for iterations. Once we have locked on our approach and method of forecasting, we can jointly replicate the same in a SQL proc or C# script. Our priority should be to get a good forecasting. I will be able to mail you an example of the analysis only by 26th evening India Time as I am travelling and unable to access my system. Thanks

      • il y a 7 ans
    2. scringhelteamsrl
      Titulaire du concours
      • il y a 7 ans

      ok. I keep in mind that this should work for many types of consumers. The tweaks should have a way of adjusting themselves accordingly with the consumer historical data.

      • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    You said "Model will be built into excel which can be integrated to the C# platform". We have historical consumption information in a MS SQL database. How would you see this integration?

    • il y a 7 ans
  • scringhelteamsrl
    Titulaire du concours
    • il y a 7 ans

    hi, thx for your entry. Please mail the forecast so we can analyse it.

    • il y a 7 ans

Montrer plus de commentaires Traitement en cours...