Ah brother again. I don't know why but I think though there are many turtorials on TH-cam I landed on your channel the reason I think is that I admire you coz you look like my age, and when a person gets a someone form whom he can learn whose age is same of his then it's a great pleasure and admiration
This is great - and I like how you go through the code. I am new to Python - transitioning from R - and understanding "why" to code something a particular way can be difficult when someone doesn't explain it. So thank you!
Madam, if you don't mind answering my query. Iam aged 24 and learning data science . I learnt R but just some basic stuff. Iam more shifted towards python coz I thought in future i can be able to switch my career from data science to some core programming and with python I learnt ML and deep learning stuff . What do you advise me on this. Is it a good thought or I should also be learning more of R.
Can you please include all variables and forecast the results with different forecasting algorithms. This one I found interesting. For multivariate forecasting can we use Arima ? I am newbie to forecasting. So any related help/ comment will be appreciated.
Great explanation thanks, what about the p-values of each of the variables in the varmax model? I don't think all of those variables are significant from what I see in the summary of model fitting step.
Hi, Nice video!!. I have a query. At 10:40, Why "train_df" is simply used in VARMAX() function instead of "train_df.df()[1:] " ? What is the role of "enfore_stationarity=True" ? does "enforce_stationarity=True" makes the model to calculate and select such a "p" value where sufficient stationarity is available ?
While using Adfuller, my Colab notebook seems to crash due to complete RAM consumption(12GB) in the basic version of Colab. Please suggest me a solution. Data size is 0.3M records, 18 columns.
Hi Nachiketa, Thanks for the nice tutorial. While trying to replicate the exact steps, I faced and error while using the VARMAX function at 11:00. { ValueError: Non-stationary starting autoregressive parameters found with `enforce_stationarity` set to True. } What could possibly be wrong? I even tried removing that optional parameter, but, of course, the same output. Another doubt: in the first cell, while importing libraries, we can ignore importing the "tqdm". Right? Thanks again.
Great video! I was wondering how you would test for white-nose residuals in the VAR vector error terms given that you got 4 lags to be optimal from AIC?
hii i am facing error while running the select_order thing...... TypeError: loop of ufunc does not support argument 0 of type float which has no callable sqrt method.............please help unable to solve this error
hey lovely video, i wanted to ask that in ad fuller test why u used print(realgdp/realcons) when i am using the same on other data set it is not working, please help.
what happens if you have one dependent time series with an integration order of 1 and the independent time series has an order of integration of 2?, how do you perform the granger test in that case?, do you have to differentiate the dependent time series one more time ?
the p-value of independent variables in VAR output seem high, is that a problem? I read some similar examples of VAR but no one actually talk about this issue
Hii, at 12:48 of the video, what is happening when we did "predictions= predict.predicted_mean" ? Which variables are included in the mean calculation ?? Can anyone explain please..
I think it's because VAR() model needs stationary data, so when you do train_df.diff()[1:], make it stationary as he has shown in adfuller test. At 11:00 he explains about VARMAX() model, which with the parameter enforce_stationarity = True, make it easier since you don't need to provide the stationary data (which afterwards you'll need to invert the transformation), cause the model already transform your input data for you. You can do the classic way using the train_df.diff() as input of regressor model, but it'll take more lines to do the same thing as he did.
@@leelasaivoonna1728 I'd ensure that the data is stationary before training the model. You can take the first difference and do ADF test to confirm that you're ok to go. And back to "why stationarity", these models like VAR and ARIMA have such a mean reversal behavior. The further your predictions go, the more it will look like a flat line. If your data isn't stationary, let's say it's incresing along time, the true data will take off to the roof while the predictions will stay in the middle, even if the short term perdictions, the generalization is affected.
![[Part 1] ทารกน้อยถูกทิ้ง👶 หญิงชราเก็บขวดขายเลี้ยงดูเขาจนโต #โลกการละคร](http://i.ytimg.com/vi/nPTDvFCExks/mqdefault.jpg)
Ah brother again. I don't know why but I think though there are many turtorials on TH-cam I landed on your channel the reason I think is that I admire you coz you look like my age, and when a person gets a someone form whom he can learn whose age is same of his then it's a great pleasure and admiration
Life saver. Your explanation and demonstration is way better than ChatGPT.
This is great - and I like how you go through the code. I am new to Python - transitioning from R - and understanding "why" to code something a particular way can be difficult when someone doesn't explain it. So thank you!
Madam, if you don't mind answering my query.
Iam aged 24 and learning data science . I learnt R but just some basic stuff. Iam more shifted towards python coz I thought in future i can be able to switch my career from data science to some core programming and with python I learnt ML and deep learning stuff . What do you advise me on this. Is it a good thought or I should also be learning more of R.
this playlist is gonna be a great help for my interview today amazing content brother
Glad it was helpful!
Can you please include all variables and forecast the results with different forecasting algorithms. This one I found interesting. For multivariate forecasting can we use Arima ? I am newbie to forecasting. So any related help/ comment will be appreciated.
Love the parrots in the background! You could teach them to say AI, AI, AI.
It was Simply Superb Bro 🙌. Also Nice Name 👍. Video with Detailed explanation of parameters would be Nice. Thank you
Excellent Brother!! Good Job you explained it very well :👏
Fantastic Video. Have a fantastic day Nachiketa
Pls some more on tutorial on multi variate forecasting ...time series. .like Dart , pycaret....thanks in advance...
Helpful, thanks for explaining each line
you're welcome :)
Thanks for the fantastic video. If I were to use all columns how would I apply granger causality?
Great explanation thanks, what about the p-values of each of the variables in the varmax model? I don't think all of those variables are significant from what I see in the summary of model fitting step.
This is very helpful. thanks for sharing the video
Hi, Nice video!!. I have a query.
At 10:40, Why "train_df" is simply used in VARMAX() function instead of "train_df.df()[1:] " ? What is the role of "enfore_stationarity=True" ?
does "enforce_stationarity=True" makes the model to calculate and select such a "p" value where sufficient stationarity is available ?
While using Adfuller, my Colab notebook seems to crash due to complete RAM consumption(12GB) in the basic version of Colab. Please suggest me a solution. Data size is 0.3M records, 18 columns.
Is it helpful for forecasting sales with multiple dependent variable?
How can I write equation/formula based on that model result?
Hi Nachiketa,
Thanks for the nice tutorial. While trying to replicate the exact steps, I faced and error while using the VARMAX function at 11:00.
{ ValueError: Non-stationary starting autoregressive parameters found with `enforce_stationarity` set to True. }
What could possibly be wrong? I even tried removing that optional parameter, but, of course, the same output.
Another doubt: in the first cell, while importing libraries, we can ignore importing the "tqdm". Right?
Thanks again.
Great video! I was wondering how you would test for white-nose residuals in the VAR vector error terms given that you got 4 lags to be optimal from AIC?
If the variables are of different scales , is it imperative to scale them all like you do for typical ML models ? Good video btw. Thanks.
Great job man
can you create a video for multivariate time series forecasting using GRU and LSTM
Thanks Buddy, great content!
hii i am facing error while running the select_order thing......
TypeError: loop of ufunc does not support argument 0 of type float which has no callable sqrt method.............please help unable to solve this error
hey lovely video, i wanted to ask that in ad fuller test why u used print(realgdp/realcons) when i am using the same on other data set it is not working, please help.
what happens if you have one dependent time series with an integration order of 1 and the independent time series has an order of integration of 2?, how do you perform the granger test in that case?, do you have to differentiate the dependent time series one more time ?
11:16 I have a question. so are the standard errors the values for all the residual error terms?
Helpful video 🎉
What If we have Week instead of date column....
Can I use parse_date= 'Week'
Great work
You checked for stationarity but you didn't use it during Granger Testing? Aren't you supposed to use the differenced series in the testing?
the p-value of independent variables in VAR output seem high, is that a problem? I read some similar examples of VAR but no one actually talk about this issue
bro in prediction my year index column convert into numbers what need. to do please help
good work
great content. very helpful
RMSE for ULC was at 54. which is high, how do we get that value down.
Hii, at 12:48 of the video, what is happening when we did "predictions= predict.predicted_mean" ? Which variables are included in the mean calculation ??
Can anyone explain please..
Excellent content !!
Brilliant bro!
Thanks, great stuff!
Thanks!
Thanks for sharing
Good....Informative
Really helpful 👍
hello at 8:28 why did you use the diff()[1:]? and not just the train_df ?
I think it's because VAR() model needs stationary data, so when you do train_df.diff()[1:], make it stationary as he has shown in adfuller test.
At 11:00 he explains about VARMAX() model, which with the parameter enforce_stationarity = True, make it easier since you don't need to provide the stationary data (which afterwards you'll need to invert the transformation), cause the model already transform your input data for you.
You can do the classic way using the train_df.diff() as input of regressor model, but it'll take more lines to do the same thing as he did.
brilliant!
In 7:20, can you explain again, does ulc cause rgnp? I don't get it.
ulc doesn't cause rgnp at lag 1, but does in lag 2, 3, and 4
Excelentes videos nuevo seguidor :)
감사합니다
Hi! Can you show how to make VAR model in Exel? Pls
5:43 so the dataset is stationary?
Thanks dude
i don't get how he got the date to show up in the predictions column
TE AMO
Hey if your doing this with non stationary data and it came with good prediction, why doing stationary is necessary??
Because of consistency. I mean, if your data isn't stationary, you can't build a wide and general model.
@@xeiquespirwilliam2167 hey but what should we do for VARMAX how to know the Q value..
@@leelasaivoonna1728 I'd ensure that the data is stationary before training the model. You can take the first difference and do ADF test to confirm that you're ok to go.
And back to "why stationarity", these models like VAR and ARIMA have such a mean reversal behavior. The further your predictions go, the more it will look like a flat line. If your data isn't stationary, let's say it's incresing along time, the true data will take off to the roof while the predictions will stay in the middle, even if the short term perdictions, the generalization is affected.
Y a ale tambien
Not 'cause', it's 'granger cause', please correct yourself