Exponential Smoothing Forecast Python and Power BI

Learn how to incorporate triple exponential smoothing forecast models in Power BI with the help of Python. Train and test your forecast with these easy to follow steps. Find me on LinkedIn

[@absentdata]

You can get the dataset and pbix code here. Please like and subscribe: github.com/Gaelim/Exponential-Forecast

[@MrFlex-fq9yt]

could you please share the ipynb file

[@zoomskiller]

Thanks for the helpful video. You're a hero for attempting to help these people in the comments who don't know how to code lol.

[@ruslanakh8381]

Amazing channel, subscribed!

[@Tanius_Sena]

Thanks for sharing ...It helped a lot here. I would like to complete by saying that it is necessary to create a virtual machine with python 3.7 and point to the PBI with the libraries and their respective versions installed.

[@haditadayyon3146]

Hi thanks for sharing this video. Could you explain how you built the time series plot in power bi that shows the actuals and forecast? When I try to plot power bi places the two time series on top of each other rather than next to each other.

[@absentdata]

Yes I can. You should have an additional table the forecast. Then append the actuals and forecast

[@phinebucad9033]

Hi, I'm a newbie here. what if I have 2 products and I want to forecast for each product, how should the code be in power query?

[@absentdata]

2 separate products is seen as two separate datasets, so there should be two separate codes

[@nachiketbhavsar1582]

hey actually i am getting an error of train not defined in my project. so since it is data set how will u define it. can u help me with that.

[@absentdata]

If something is not defined, it means that that variable doesnt exist. Check you code

[@hatemhamdi9455]

hi sir, how did you change the type of passengers from text to number ? it always gives me an error

[@absentdata]

Make sure you don't have a string in the column. You can filter it to see that in Excel or Power BI to ensure that there is not some odd value where numbers should be

[@tlholowaka]

Hi when I use predictions = model.forecast(36) in power BI I get this error:TypeError: unsupported operand type(s) for +: 'int' and 'str'

[@absentdata]

You have a data type issue. Read the error. It indicates that you are mixing strings and integers. Check StackOverFlow for your specific case.

[@absentdata]

Read the error. Its saying that you have mixed data types

[@florianbachle4234]

@@absentdata Hey, got the same error, but don't know ho to fix. Can you advise me. Furthermore i got a error type in jupiter note: ValueError: Buffer dtype mismatch, expected 'double' but got 'long long' for model = ExponentialSmoothing(train['#Passengers'],trend='mul',seasonal='mul',seasonal_periods=12).fit()

[@tlholowaka]

@@florianbachle4234 I changed this line and it fixed the error: dataset.set_index('Month',inplace=True)

[@zoomskiller]

You need to cast the string to an int, like this int(stringValue)

[@jordiperrilliat8079]

When I use model.forecast.plot() it shows an error: AttributeError: 'numpy.ndarray' object has no attribute 'plot

[@absentdata]

Well this means that you have changed your data frame into a numpy array. You need to have a data frame to use plot

[@asifrehman4214]

Dear Sir I am not been able to run the code that runs fine on Spyder but not on PBI Script. If you can help or elaborate it further. I am a newbie towards python.

[@absentdata]

I have no idea what your configuration is or what error is being reported so it would be your responsibility to elaborate.

[@asifrehman4214]

@@absentdata Dear Sir apologizing for the silly question When I ran the python code in PBIDesktop probably the ExponenetialSmoothing firing error as below: File "statsmodels\tsa\_exponential_smoothers.pyx", line 245, in statsmodels.tsa._exponential_smoothers._holt_win_mul_mul_dam ValueError: Buffer dtype mismatch, expected 'double' but got 'long long' I have changed the data type in the following line after #Passenger column model=ExponentialSmoothing(dataset['#Passengers'].astype(n.double),trend='mul',seasonal='mul',seasonal_periods=12).fit() So it ran without any error but stuck at the following predictions line with the error as below range=p.date_range('01-01-1961',periods=36, freq='MS') predictions=model.forecast(36) TypeError: unsupported operand type(s) for +: 'int' and 'str' If you can help plz.

[@absentdata]

@@asifrehman4214 This issue is that the data type you are trying to run the prediction on is a string and not a integer. You need to ensure that the datatypes are numerical in the dataset also so chekc your Power BI data types in the Power Query Editor

[@miriraimunashemapanda4317]

Where can l get the dataset you used? I would love to do the tutorial using the same dataset

[@absentdata]

Hi you can find it here www.absentdata.com/power-bi/forecasting-with-python-and-power-bi/

[@asifrehman4214]

Dear Sir , The code run perfectly in python (spyder) but it fires error when I ran the same code in Power BI python script on the model line as below: model=ExponentialSmoothing(dataset['#Passengers'],trend='mul',seasonal='mul',seasonal_periods=12).fit() with the error code as ErrorCode=-2147467259 ExceptionType=Microsoft.PowerBI.Scripting.Python.Exceptions.PythonScriptRuntimeException If you can help.

[@asifrehman4214]

this was due to the data type and i have resolved it using dataset=dataset.astype('double') Done. Good tutorial.

[@MxxxMusic]

sir do you have discord? please help me i cant continue after 3:35 / exponential smoothing phase

[@NovusProduction1]

hi sir. is there any video that you explain how to make the code in python about this topic (exponential smoothing) in power BI for a beginner like me ? i really need it :((