It would have been interesting to run the prediction for a historic time period to compare the prediction with what actually happened. That's how we usually test our models!
brother i want to build the same project but i want to provide my data set in csv format is it possible and if it is possible what changes should i make?
Could you please assist me in resolving this issue below? I'm attempting to load the neuralprophet trained model from a different device.encountering a"Error": "[WinError 5] Access is denied:'model trained directory name ' " on model.predict(df) line. How do I figure this out...?
I’m confused on what data the model is using to predict the “actual prediction”. If it is predicting the only dataset that you gave it, what is it using to make those predictions
I have split the dataset such that I am only providing 80 percent of the actual data and the rest 20 is what the machine hasn't seen yet and we make it predict on these dates and compare with the actual values
Definitely! Infact, we just released a video on actually performing background removal of images using python's scikit learn....This is very similar and you might like it very well! I am SO GLAD you like my teaching style! here's is the link:- th-cam.com/video/KeIP7tz-33U/w-d-xo.html
Thank you! I'm glad I could confuse you just enough to make it seem like you learned something. 😉 anyways, you might like our image processing video which is kind of similar to this video when you see here's the link:- th-cam.com/video/KeIP7tz-33U/w-d-xo.html
This is great work and very interesting. I did notice though when the predicted graph was created its too small, making the X axis' values run into one another, which makes the graph inaccurate. Is there a way to make the graph bigger? I tried various MatPlotLib functions such as set_figwidth, tight, and others but nothing will make the graph larger. Thank you!
Sure there is, You can actually tilt the words by an angle, giving it more room for existance. Additionally, I am sure there is a function to only display xvalues at regular intervals I recommend you check on google or directly ask chatgpt
You bet! 😂 Anyways, don't consider this a promotion but I am so sure you are going to like the video talking about building a language translator. Do check it out th-cam.com/video/OwA8mszL38w/w-d-xo.html
Well, actually this is just a pretrained model We just get the weights on to our local machine and then use .predict function This way of using a pretrained model mostly works in any project Again, code file in the description Cheers
@@SMDS_Studioyes i know but if we had excell sheet with all numbers 99 results can we predict the next winning lottery winning numbers with machine learning or data analysis or some AI statistics or mathematics !? Can we try it bro just for fun I will send you the excel!?
HI, I enjoyed the code. I tried running your script but I'm getting a compatibility error between Python latest version and NeuralProphet. What version of Python are you using?
Hi! I am glad you enjoyed the code. Python 3.9 was used in this video. I am not sure if your python version is updated to the latest and I recommend trying to upgrade it to 3.9 or higher, as NeuralProphet may have updates that align better with newer Python releases. Let me know if that helps!
Two questions: 1. You trained the model on the training data, then used it to predict the training data, so naturally it will look accurate. What was the idea or purpose behind this? I am genuinely just curious to see if I am missing something 2. The model forecasts for weekends too and Saturday and Sunday are included in time series. Naturally, stock's do not trade on weekends. Can this be adapted in the model? Perhaps the model can only identify existing trends?
Let's answer this 1 by 1 1.) So I took the first 80 percent of the data and gave it to the model for training. Now it doesn't know anything about the rest 20 percent. Then I made predictions with the model for the duration of which we know the actual values but then the model doesn't Then I plotted both actual and predicted values to compare 2.) Here, I just use the forecast function which will give predictions continuously for the mentioned days So this goes to say, that the project itself is just a demonstration of how the prophet model Where as when it comes to real world stocks project, there are many other factors we have to consider for modeling We have planned a sequel video on that....soon
@@SMDS_Studio this helped me a lot with time series forecasting. I'm curious when you split the model 80/20 though? I don't see that step in the video. Thanks for helping me out!
I ran your code in Pycharm and it failed to run at the line: stocks = pd.read_csv('stock_data_Samsung.csv') unless I change it to stocks: DataFrame = pd.read_csv('stock_data_Samsung.csv')
How did you make it to work? It doesn't install in anaconda for jupyter notebook. It installs in google colab, but when I tries to train it, it reset colab processes.
As I mentioned in the video, the model itself is very unreliable and by no means this would be a financial advice Ofc you can use the bitcoin data to see predictions but again I wanna say that this is just a representation of how a prophet model works
I want to add the holidays dataframe and include a few regressors for the model. Do you know how to add them??
It would have been interesting to run the prediction for a historic time period to compare the prediction with what actually happened. That's how we usually test our models!
Interesting question here, can u analyse volume patterns to predict pump and dump schemes.
Ig you could try to use FBPROPHET for that but I am not really sure
Although, it might work..
brother i want to build the same project but i want to provide my data set in csv format is it possible and if it is possible what changes should i make?
Congratulations for your excellent video. Never heard about neuralprophet library, and will give it a try.
Thank you !
Glad it was helpful!
Great video bro! Subscribed!
Awesome! Thank you!
Could you please assist me in resolving this issue below?
I'm attempting to load the neuralprophet trained model from a different device.encountering a"Error": "[WinError 5] Access is denied:'model trained directory name ' " on model.predict(df) line.
How do I figure this out...?
Are you enccountering this in google collab?
I’m confused on what data the model is using to predict the “actual prediction”. If it is predicting the only dataset that you gave it, what is it using to make those predictions
I have split the dataset such that I am only providing 80 percent of the actual data and the rest 20 is what the machine hasn't seen yet and we make it predict on these dates and compare with the actual values
@@SMDS_Studiowhere exactly are you splitting it into 80 and 20 because i dont see that in the code. Is that what the model.fit() does for you?
If you have a look closely, I have removed the recent 20 percent dates and provided only the first 80 percent
Using slicing
@@SMDS_Studiowhere do you do that at? I’m not seeing it in the code. Unless that is what model.fit(Stocks) does?
Thank you very interesting.
Very consice, thank you a lot!
Glad it was helpful!
More videos on this thank you i love your teaching , you are very good in teaching
Definitely! Infact, we just released a video on actually performing background removal of images using python's scikit learn....This is very similar and you might like it very well! I am SO GLAD you like my teaching style!
here's is the link:- th-cam.com/video/KeIP7tz-33U/w-d-xo.html
Thank you, good tutorial
Thank you! I'm glad I could confuse you just enough to make it seem like you learned something. 😉
anyways, you might like our image processing video which is kind of similar to this video when you see
here's the link:- th-cam.com/video/KeIP7tz-33U/w-d-xo.html
what is the programme he is writing in - i am using cmd
Jupyter notebook
This is great work and very interesting. I did notice though when the predicted graph was created its too small, making the X axis' values run into one another, which makes the graph inaccurate. Is there a way to make the graph bigger? I tried various MatPlotLib functions such as set_figwidth, tight, and others but nothing will make the graph larger. Thank you!
Sure there is,
You can actually tilt the words by an angle, giving it more room for existance.
Additionally, I am sure there is a function to only display xvalues at regular intervals
I recommend you check on google or directly ask chatgpt
Great topic, thanks 👍
My pleasure!
In fact, we are working on similar type videos that are easier to replicate as well
Hope you stay tuned!
Best video ❤
So nice to get the acknowledgement!
We would be much much delighted if you decide to spend more time watching our videos 😉
This is so exciting mate. Thanks!
You bet! 😂
Anyways, don't consider this a promotion but I am so sure you are going to like the video talking about building a language translator. Do check it out
th-cam.com/video/OwA8mszL38w/w-d-xo.html
I'm trying to make a machine learning model for creating an automated forex trading strategy.
Curious if this transferable?
Well, actually this is just a pretrained model
We just get the weights on to our local machine and then use .predict function
This way of using a pretrained model mostly works in any project
Again, code file in the description
Cheers
Can we do it to predict winning lottery number
It might not be very accurate
As you know lottery is very very very vague
@@SMDS_Studioyes i know but if we had excell sheet with all numbers 99 results can we predict the next winning lottery winning numbers with machine learning or data analysis or some AI statistics or mathematics !? Can we try it bro just for fun I will send you the excel!?
HI, I enjoyed the code. I tried running your script but I'm getting a compatibility error between Python latest version and NeuralProphet. What version of Python are you using?
Hi! I am glad you enjoyed the code. Python 3.9 was used in this video. I am not sure if your python version is updated to the latest and I recommend trying to upgrade it to 3.9 or higher, as NeuralProphet may have updates that align better with newer Python releases. Let me know if that helps!
Two questions:
1. You trained the model on the training data, then used it to predict the training data, so naturally it will look accurate. What was the idea or purpose behind this? I am genuinely just curious to see if I am missing something
2. The model forecasts for weekends too and Saturday and Sunday are included in time series. Naturally, stock's do not trade on weekends. Can this be adapted in the model? Perhaps the model can only identify existing trends?
Let's answer this 1 by 1
1.) So I took the first 80 percent of the data and gave it to the model for training.
Now it doesn't know anything about the rest 20 percent. Then I made predictions with the model for the duration of which we know the actual values but then the model doesn't
Then I plotted both actual and predicted values to compare
2.) Here, I just use the forecast function which will give predictions continuously for the mentioned days
So this goes to say, that the project itself is just a demonstration of how the prophet model
Where as when it comes to real world stocks project, there are many other factors we have to consider for modeling
We have planned a sequel video on that....soon
@@SMDS_Studio this helped me a lot with time series forecasting. I'm curious when you split the model 80/20 though? I don't see that step in the video. Thanks for helping me out!
That's awesome 🎉
I value your acknowledgement ;)
In fact, I believe you would like our latest video as well
Here's the link:- th-cam.com/video/FP10wx_yP3A/w-d-xo.html
excellent. Thank you!
SO happy that you liked this video! well continue to post such type of videos upto your expectations
thanks bro 👋
Welcome 👍
I ran your code in Pycharm and it failed to run at the line:
stocks = pd.read_csv('stock_data_Samsung.csv')
unless I change it to
stocks: DataFrame = pd.read_csv('stock_data_Samsung.csv')
This I am not really sure, I mean it should be working in the first case itself but then I recommend you to have a look at your python version
How did you make it to work?
It doesn't install in anaconda for jupyter notebook.
It installs in google colab, but when I tries to train it, it reset colab processes.
I recommend you taking a look at the python version you are using..
NeuralProphet works best on python versions 3.9 or above
Let me know if it helps
yes ,i cant run python 3.6 & python 3.10 but can run python 3.9
Great.
Appreciate your response 😃. Thanks!!
By the way, please provide code listing if you can!
My Bad, Here you go:-
github.com/SMDS-Studio/Predict-Trends-Code-File/blob/main/NeuralProphet.ipynb
subscribed..
Amazing! Ty 😊
can you try this for bitcoin please
As I mentioned in the video, the model itself is very unreliable and by no means this would be a financial advice
Ofc you can use the bitcoin data to see predictions but again I wanna say that this is just a representation of how a prophet model works
lol and here is where you lose all your money.
*video ends*
🤣
😂Think of a Machine giving a reply to this