"🔥Caltech Post Graduate Program In AI And Machine Learning - www.simplilearn.com/artificial-intelligence-masters-program-training-course?WkFhVq9-nc&Comments&RUclips 🔥IITK - Professional Certificate Course in Generative AI and Machine Learning (India Only) - www.simplilearn.com/iitk-professional-certificate-course-ai-machine-learning?WkFhVq9-nc&Comments&RUclips 🔥Purdue - Post Graduate Program in AI and Machine Learning - www.simplilearn.com/pgp-ai-machine-learning-certification-training-course?WkFhVq9-nc&Comments&RUclips 🔥IITG - Professional Certificate Program in Generative AI and Machine Learning (India Only) - www.simplilearn.com/iitg-generative-ai-machine-learning-program?WkFhVq9-nc&Comments&RUclips 🔥Caltech - AI & Machine Learning Bootcamp (US Only) - www.simplilearn.com/ai-machine-learning-bootcamp?WkFhVq9-nc&Comments&RUclips"
We are delighted to have been a part of your learning journey! If you want to continue honing your skills and keeping up-to-date with industry trends, check out our course offerings in the description box.
Machine Learning is the Future and yours can begin today. Comment below with you email to get our latest Machine Learning Career Guide. Let your journey begin
Hi, thanks for watching our video. We have sent the requested dataset to your mail ID. Do show your love by subscribing to our channel using this link: ruclips.net/user/Simplilearn and don't forget to hit the like button as well. Cheers!
Thanks for the great tutorial! A couple of questions please. To establish a Multivariate Multi-Step LSTM Models - Multiple Input Multi-Step Output: 1. How can I modify this code to take, for instance, 3 inputs to forecast a different single output that depends on those 3 inputs? 2. How can I forecast multiple timesteps in the future without knowing the new inputs (because they're in the future)?
Do you have any questions on this topic? Please share your feedback in the comment section below and we'll have our experts answer it for you. Also, if you would like to have the dataset for implementing the use case shown in the video, please comment below and we will get back to you. Thanks watching the video. Cheers !!
Hello Ankush, thanks for watching our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Hello Ankush, thanks for sharing your email ID. We have sent the requested dataset to your mail ID. Do subscribe and stay tuned for updates on our channel. Cheers :)
Hey Albert, thank you for appreciating our work. We are glad to have helped. Do check out our other tutorial videos and subscribe to us by clicking this link ruclips.net/user/Simplilearn to stay connected. Cheers :)
Congratulations for the Recurrent Neural Network (RNN) Tutorial. Then, I would like that you can help me to implement a code with LSTM to predict wind speed 6 hours ahead. The first 550 table lines would be for training and the remaining 194 for forecasting. I greatly appreciate the support. Thank you very much. Mr Pedro Jr
Hi Melaku, we are glad you found our video helpful. Do show your love by subscribing to our channel using this link: ruclips.net/user/Simplilearn and don't forget to hit the like button as well. Cheers!
if you wanted to use multiple columns from the dataset (for example, the "Open" price AND the "High" price) how would you reconfigure this code to allow for that?
If you want to add more independent variables to your model, you need to modify the following lines of code: training_set=dataset_train.iloc[:, "the number of independent columns"].values
@@SimplilearnOfficial Does simply updating iloc to a different range of columns make the net train even if you only look at open? At 37:00 you say you only look at open. If you want to train using other columns containing things like technical indicators, do I need to update that 0? Or is it a different line that needs updating
Hello I want to compare the results from RNN and LSTM, as it is explained LSTM doesn't have vanishing gradient problem. Can I use simple RNN here for the same prediction? I could expect that it is not going to be as good as LSTM. but, I was wondering how to check the same tutorial only with simple RNN. Any suggestion will be helpful.
Hi It was really useful. Really thanks. I have a question: I set this up (changed the parameters and trained). now, how can I use it in realtime? I mean, for example, I connected this program to realtime data and I receive data each one hour. should I train (fit) the network with every row of data I receive???!!! or just load the trained network at the beginning of the program and just get the predictions? in which points I should load and save the network? and after a while that new data is imported to the network, should it be retrained? please give me some advice about this. Regards
Hello Mohsen, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Hi, we are glad you enjoyed our video. We have more videos related to neural network on this playlist given in the link - ruclips.net/video/FbxTVRfQFuI/видео.html. We hope you will join our community! Happy learning!
Hey, thank you for appreciating our work. We are glad to have helped. Do check out our other tutorial videos and subscribe to us to stay connected. Cheers :)
Hi, Thank you for the helpful tutorial. Can we fit/predict the model without reshape the X_train and X_test? X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1 )) X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1)) What are the purpose of reshape? Can we skip them? Best
Hi Seena, we regret to say that we cannot help with individual needs. However, if you have any query with the video, you can post it in the comment section. Thanks.
Hello, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we can send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that too. Hope that helps.
Hi Diya, thank you for watching our video and for the honest feedback. We will definitely look into this. Do subscribe, like and share to stay connected with us. Cheers :)
Hello Vinay Kumar, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Hi Kartik, it's true that traditional RNN cannot deal with long time lags between relevant signals. We need LSTMs to solve this issue because they have the capability to memorize previous information. Hope that helps!
Hi! Very useful lecture on LSTM. I have a fundamental question: If my features are more than a simple close value, and I have cycles besides single time step, how can I modify the shape of the net in order to get a prediction? In my case I have, in each cycle, 5 features; than I must predict a remaining useful life of a device without any "run until failure data" in my dataset, but I know a certain level of threshold in which the characteristics of the device are bad. Can you help me?
Excellent tutorial and I really liked it .But I have confused between C(t-1) and h(t-1) . C(t-1) is Cell state and h(t-1) is output of first cell right? Do we use the both in the next timestep ?Because I found only h(t-1) is used in the 2nd timestep and C(t-1) hasn't been used .Or have we added the cell state t-1 with t ?
I wanted to create lstm network for classification purpose. My dataset has 30 classes, so instead of creating just one neuron in final layer by regressor.add(Dense(units=1)) , I wrote regressor.add(Dense(units=30, activation='softmax')). But when I fit the data into the model the predicted output I am obtaining is vector of size 30 but that is not my requirement.I want the output to be a single value between 1 to 30 since there are 30 classes. Please help, how do I change my code.
Hello! So in this video, we end up getting predictions that test against our test data, but is there a way to get future predicts that we don't have test data yet for?
Hello Sriram, thanks for viewing our tutorial and we hope it is helpful. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly.
Also, you have mentioned that batch size only pertains to the number of items that would be loaded into the Ram but at other places I am seeing that page size actually defines how much data the neural network could see in one shot. Although I must confess I don't understand the Latter definition
Hi, We cannot pass the entire data set into the neural network for training at once. So, we divide the dataset into number of batches. The batch size we have considered is 32. So, suppose we have a 1000 images and a batch size of 32, then an epoch would run 1000/32 = 31 iterations or steps. Hope this answered your question. Please let us know if in case you have any other queries. Thanks for watching the video. Cheers!!
Hi Shantam, thanks for checking out our tutorial. The model built in the video predicts the prices of Google stocks in 2017, based on the prices of Google stocks for the past 5 years (2012-2016). The training data contains prices of stocks from 2012-2016, while the test data contain the prices of stocks for the year 2017.
hey i need some guidelines to build a chatbot using LSTMS which is job specific like doing certain task ! so what APIs should i have to use to built one from scratch
First, you need to convert your questions and answers pairs into vectorized formats. Then, develop a Deep Learning model and validate the data. Please refer to the following link to learn more on this: towardsdatascience.com/building-an-ai-chat-bot-e3a05aa3e75f medium.com/botsupply/generative-model-chatbots-e422ab08461e "
Thank you so much for this video that cleared my brain that was about to blow trying to understand RNN and LSTM application in Data analysis. Is there any video where RNN-LSTM was used on KDDCup99 dataset for anomaly detection
Hi Imran, We don't have any relevant tutorial that shows how you tune LSTM hyperparameters with Keras. Here is a blog that can help you out in this machinelearningmastery.com/tune-lstm-hyperparameters-keras-time-series-forecasting.
Thank you for an excellent tutorial and also for using a practical real life example. I was going round in circles trying to figure out what was the last parameter as one when we were reshaping in numpy array and now I understand it is just an indicator that the number of parameters is now ending. In this example you have used three layers of lstm ? Is there any information on how the learning process and also the outcome would vary when we used different layers of Lstm
Hi, There are 4 Lstm layers used while training the RNN model. To be fair, there is no proper way to determine a network topology just from the number of inputs and outputs. It depends on the number of training samples and the complexity of the model you are trying to train. In RNN, 4-5 LSTM layers will generally produce a good model. You will find a variation in the output to a very small extent. Hope this answered your question. Please let us know if in case you have any other queries. Thanks for watching the video. Cheers!!
Hello sir after running the below command # Getting the prediction stock price of 2017 dataset_total= pd.concat((dataset_train['Open'], dataset_test['Open']), axis=0) inputs = dataset_total[len(dataset_total) - len(dataset_test) - 60:].values inputs= inputs.reshape(-1,1) inputs= sc.transform(inputs) x_test =[] for i in range(60, 80): x_test.append(inputs[i-60:i, 0]) x_test = np.array(x_test) x_test= np.reshape(x_test, (x_test.shape[0], x_test.shape[1],1)) predicted_stock_price= regressor.predict(x_test) predicted_stock_price= sc.inverse_transform(predicted_stock_price) i got this error ### ValueError: Found array with dim 3. None expected
I have used this for weather prediction. The predicted value of training and testing when plotted against the actual value are seems to be shifted by 1 step, can you explain the reason.
Hi Imran, Here are the steps you can try to solve this issue: 1. Decrease the LSTM dimension 2. Add some form of regularization to combat overfitting (dropout) 3. Train for more epochs or change the learning rate.
@@SimplilearnOfficial I have tried all the steps that you have mentioned. Can you give me some idea how to decide LSTM dimension. I mean from which size to start. For eg. In my case I have 15000 rows and 8 coloumns in training dataset. So from how many units should i start in LSTM layer and also how many number of layers. Thankyou.
In most of my cases it happens that loss during training keeps on decreasing but validation loss is changing between some fixed range. Any suggestions for this??? Thankyou.
"Hi Imran, This happens due to improper hyper-parameters used while training. Please adjust the hyper-parameters and train the model for more number of epochs."
"Hi Imran, Check out this link for choosing the right hyperparameters for LSTM towardsdatascience.com/choosing-the-right-hyperparameters-for-a-simple-lstm-using-keras-f8e9ed76f046"
I am getting the error numpy.ndarray' object has no attribute 'append' when I run the command X_train.append(training_set_scaled[i-60:i, 0])...Any idea how to solve this ?
"Hi Devesh, Please refer to the following link that will help you solve this issue: stackoverflow.com/questions/42942781/python-attributeerror-numpy-ndarray-object-has-no-attribute-append/"
@@SimplilearnOfficial If we consider that neighborhood pixels are similar then a small patch of image can be classified using LSTM.It is implemented in an IEEE paper.
Thanks for your input. It is true that LSTMs can be used for general image classification problems. But not for implementing any major visual imagery task.
Hey Oz, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we can send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that too. Hope that helps.
Hi, thanks for watching our video. We have sent the requested dataset to your mail ID. Do show your love by subscribing to our channel using this link: ruclips.net/user/Simplilearn and don't forget to hit the like button as well. Cheers!
Hello Ankush, thanks for watching our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Hi, thanks for watching our video. We have sent the requested dataset to your mail ID. Like and share our video with your peers and also do not forget to subscribe to our channel for not missing video updates. We will be coming up with more such videos. Cheers!
Thank you for the great video! I have a problem however with the real forecast of the model. Looks like it can be only use to compare the already existing data, but is very difficult to make it predict non existing values. Can you also make a video about that part? Should the model be saved and after that used may be?
Hello, thank you for watching our video. We will definitely look into your suggestions. Do subscribe and stay tuned for updates on our channel. Cheers :)
"Karas" and "num-pee"... I honestly couldn't tell whether or not this guy was trolling everyone on the internet... Good video. though For everyone else: it's "Keras" (you wont find anything related to "Karas"). Additionally, "numpy" is pronounced "num-pie", just like "python" ("pie-thon")... unless you pronounce it as "peethon" and you like to fondle/"Karas" sweaters... :p great video though... were you trolling the internet? I seriously can't tell.
No, it must have been the accent issue. We definitely share the feedback to the concerned team and make sure it is not happening again. Thanks for your support and love!
Hello Nick, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Thank you for your review. We are sorry to hear you had such a frustrating experience, but we really appreciate you bringing this issue to my attention
Hey Aref, thank you for watching our video and for the honest feedback. We will definitely look into this. Do subscribe, like and share to stay connected with us. Cheers :)
Hello Ravi, thanks for viewing our tutorial and we hope it is helpful. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly.
Hey Reddy, thank you for appreciating our work. We are glad to have helped. Do check out our other tutorial videos and subscribe to us to stay connected. Cheers :)
Hello, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
"🔥Caltech Post Graduate Program In AI And Machine Learning - www.simplilearn.com/artificial-intelligence-masters-program-training-course?WkFhVq9-nc&Comments&RUclips
🔥IITK - Professional Certificate Course in Generative AI and Machine Learning (India Only) - www.simplilearn.com/iitk-professional-certificate-course-ai-machine-learning?WkFhVq9-nc&Comments&RUclips
🔥Purdue - Post Graduate Program in AI and Machine Learning - www.simplilearn.com/pgp-ai-machine-learning-certification-training-course?WkFhVq9-nc&Comments&RUclips
🔥IITG - Professional Certificate Program in Generative AI and Machine Learning (India Only) - www.simplilearn.com/iitg-generative-ai-machine-learning-program?WkFhVq9-nc&Comments&RUclips
🔥Caltech - AI & Machine Learning Bootcamp (US Only) - www.simplilearn.com/ai-machine-learning-bootcamp?WkFhVq9-nc&Comments&RUclips"
awesome content and explanation. Adding a sense of humour while teaching for 1 hour non-stop is excellent. Good job sir
We are delighted to have been a part of your learning journey! If you want to continue honing your skills and keeping up-to-date with industry trends, check out our course offerings in the description box.
Machine Learning is the Future and yours can begin today. Comment below with you email to get our latest Machine Learning Career Guide. Let your journey begin
oasishaon@gmail.com
Hi, thanks for watching our video. We have sent the requested dataset to your mail ID. Do show your love by subscribing to our channel using this link: ruclips.net/user/Simplilearn and don't forget to hit the like button as well. Cheers!
mishal.morani191@gmail.com
Could you please send dataset to my mail
swasthiksomayaji4@gmail.com
Could you please send dataset to my mail
wjson4240@gmail.com
Thanks for the great tutorial! A couple of questions please.
To establish a Multivariate Multi-Step LSTM Models - Multiple Input Multi-Step Output:
1. How can I modify this code to take, for instance, 3 inputs to forecast a different single output that depends on those 3 inputs?
2. How can I forecast multiple timesteps in the future without knowing the new inputs (because they're in the future)?
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Do you have any questions on this topic? Please share your feedback in the comment section below and we'll have our experts answer it for you. Also, if you would like to have the dataset for implementing the use case shown in the video, please comment below and we will get back to you. Thanks watching the video. Cheers !!
Hi thanks for this great tutorial. Where can I find the dataset and the python notebook?
Please send dataset.
Hello Ankush, thanks for watching our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Hello Ankush, thanks for sharing your email ID. We have sent the requested dataset to your mail ID. Do subscribe and stay tuned for updates on our channel. Cheers :)
I would like to get the data set. Please send dataset
Realy cool tutorial on rnn. Thank you for your work😍😍
You're welcome 😊
Excellent video. Thank you for offering it. at 12:45 discussing vanishing while slide is exploding...timing off.
Hey Albert, thank you for appreciating our work. We are glad to have helped. Do check out our other tutorial videos and subscribe to us by clicking this link ruclips.net/user/Simplilearn to stay connected. Cheers :)
awesome video ! Just a hint, in 32:50 of the video you could just do: dataset_train[['Open','High']].values
Hi Marcos, thanks for your kind comment. We acknowledge your alternate suggestion as well and you are right about it. Kudos!
Congratulations for the Recurrent Neural Network (RNN) Tutorial. Then, I would like that you can help me to implement a code with LSTM to predict wind speed 6 hours ahead. The first 550 table lines would be for training and the remaining 194 for forecasting. I greatly appreciate the support. Thank you very much. Mr Pedro Jr
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Excellent talk!!! Thank you.
You are so welcome!
Thanks for your much helpful demonstration
Hi Melaku, we are glad you found our video helpful. Do show your love by subscribing to our channel using this link: ruclips.net/user/Simplilearn and don't forget to hit the like button as well. Cheers!
Very well explained, thanks
Hi Diego, glad that you found our video helpful. Do support us by liking, sharing and subscribing to our channel! Cheers :)
if you wanted to use multiple columns from the dataset (for example, the "Open" price AND the "High" price) how would you reconfigure this code to allow for that?
If you want to add more independent variables to your model, you need to modify the following lines of code: training_set=dataset_train.iloc[:, "the number of independent columns"].values
@@SimplilearnOfficial Does simply updating iloc to a different range of columns make the net train even if you only look at open? At 37:00 you say you only look at open. If you want to train using other columns containing things like technical indicators, do I need to update that 0? Or is it a different line that needs updating
Hello I want to compare the results from RNN and LSTM, as it is explained LSTM doesn't have vanishing gradient problem. Can I use simple RNN here for the same prediction? I could expect that it is not going to be as good as LSTM. but, I was wondering how to check the same tutorial
only with simple RNN. Any suggestion will be helpful.
Hi
It was really useful. Really thanks.
I have a question:
I set this up (changed the parameters and trained).
now, how can I use it in realtime? I mean, for example, I connected this program to realtime data and I receive data each one hour. should I train (fit) the network with every row of data I receive???!!! or just load the trained network at the beginning of the program and just get the predictions?
in which points I should load and save the network? and after a while that new data is imported to the network, should it be retrained?
please give me some advice about this.
Regards
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Hi
Thank you for this great video,
Can give the dataset to implement the use case?
Hello Mohsen, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
@@SimplilearnOfficial How I can send you my email? Here?
Hi Mohsen, you can share your email ID with no hesitation coz no comments get published without our approval.
Great Video, Very well explained.
Hi, we are glad you enjoyed our video. We have more videos related to neural network on this playlist given in the link - ruclips.net/video/FbxTVRfQFuI/видео.html. We hope you will join our community! Happy learning!
Excellent Explanation, Thank you so much for this video , Is there any video regarding text generation using LSTM _ RNN
Hey, thank you for appreciating our work. We are glad to have helped. Do check out our other tutorial videos and subscribe to us to stay connected. Cheers :)
Hey! I did it!
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Hi, Thank you for the helpful tutorial. Can we fit/predict the model without reshape the X_train and X_test?
X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1 ))
X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))
What are the purpose of reshape? Can we skip them? Best
very good thank you very much
Hey Mehran, thank you for watching our video. We are glad that you liked our video. Do subscribe and stay connected with us. Cheers :)
thank you its good video please help me i went to use RNN for text classification and how its work
Hi Seena, we regret to say that we cannot help with individual needs. However, if you have any query with the video, you can post it in the comment section. Thanks.
Pretty cool!!!
Where can I get the dataset?
Thank you in advance!
Hello, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we can send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that too. Hope that helps.
please provide a tutorial for multivariate inputs also
Hi Diya, thank you for watching our video and for the honest feedback. We will definitely look into this. Do subscribe, like and share to stay connected with us. Cheers :)
How did you took range limit as 1258 in for loop? plz give quick reply
That would be the upper bound of the training set. I guess there are 1258 entries in the training set.
in the formula for f_t in step 1, does [h, x] denote a Lie Bracket?
Thanks for the Video. Can you please share the dataset used.
Hello Vinay Kumar, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Excellent explanation. It is possible to build a recurrent neural network to use it in forex, based on indicators: MACD, cci, RSI, bb, Ichimoku?
well nice explanation but in stock section there is a problem called "time lag" which returns last close values from the given sequence.
Hi Kartik, it's true that traditional RNN cannot deal with long time lags between relevant signals. We need LSTMs to solve this issue because they have the capability to memorize previous information. Hope that helps!
why is the regressor.fit function not showing "accuracy"
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Hi! Very useful lecture on LSTM. I have a fundamental question: If my features are more than a simple close value, and I have cycles besides single time step, how can I modify the shape of the net in order to get a prediction?
In my case I have, in each cycle, 5 features; than I must predict a remaining useful life of a device without any "run until failure data" in my dataset, but I know a certain level of threshold in which the characteristics of the device are bad. Can you help me?
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Excellent tutorial and I really liked it .But I have confused between C(t-1) and h(t-1) . C(t-1) is Cell state and h(t-1) is output of first cell right? Do we use the both in the next timestep ?Because I found only h(t-1) is used in the 2nd timestep and C(t-1) hasn't been used .Or have we added the cell state t-1 with t ?
Thanks for watching our video and sharing your thoughts. Do subscribe to our channel and stay tuned for more. Cheers!
Hi..can i use this technique to analyse my multivariate time series data in econometrics?
difficult to read the code with that font size in the full screen record
Thank you so much for bringing this to our attention. We reported this right away to the relevant department.
I wanted to create lstm network for classification purpose.
My dataset has 30 classes, so instead of creating just one neuron in final layer by regressor.add(Dense(units=1)) , I wrote regressor.add(Dense(units=30, activation='softmax')).
But when I fit the data into the model the predicted output I am obtaining is vector of size 30 but that is not my requirement.I want the output to be a single value between 1 to 30 since there are 30 classes.
Please help, how do I change my code.
Hello! So in this video, we end up getting predictions that test against our test data, but is there a way to get future predicts that we don't have test data yet for?
Definitely, it is possible to predict the future.
@@SimplilearnOfficial can I have the code please for getting predictions for future
Hello Sriram, thanks for viewing our tutorial and we hope it is helpful. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly.
can i use LSTM for energy prediction in house , if humidity and temp kind of datasets are given???
Yes, it is possible to predict energy usage using LSTM.
Also, you have mentioned that batch size only pertains to the number of items that would be loaded into the Ram but at other places I am seeing that page size actually defines how much data the neural network could see in one shot. Although I must confess I don't understand the Latter definition
Hi, We cannot pass the entire data set into the neural network for training at once. So, we divide the dataset into number of batches. The batch size we have considered is 32. So, suppose we have a 1000 images and a batch size of 32, then an epoch would run 1000/32 = 31 iterations or steps. Hope this answered your question. Please let us know if in case you have any other queries. Thanks for watching the video. Cheers!!
what exactly does the model outputs when you input the test data?
Hi Shantam, thanks for checking out our tutorial. The model built in the video predicts the prices of Google stocks in 2017, based on the prices of Google stocks for the past 5 years (2012-2016). The training data contains prices of stocks from 2012-2016, while the test data contain the prices of stocks for the year 2017.
hey i need some guidelines to build a chatbot using LSTMS which is job specific like doing certain task ! so what APIs should i have to use to built one from scratch
First, you need to convert your questions and answers pairs into vectorized formats.
Then, develop a Deep Learning model and validate the data.
Please refer to the following link to learn more on this:
towardsdatascience.com/building-an-ai-chat-bot-e3a05aa3e75f
medium.com/botsupply/generative-model-chatbots-e422ab08461e
"
Thank you so much for this video that cleared my brain that was about to blow trying to understand RNN and LSTM application in Data analysis.
Is there any video where RNN-LSTM was used on KDDCup99 dataset for anomaly detection
We are glad that you liked our video. We don't have any video on anomaly detection using KDDCup99 dataset on our RUclips channel.
Didn't you use the same test data-set in stock price prediction as real stock price?
He use the... Training data set to train the model. And use it on test data set to to find the accuracy of the model
Thanks for the input!
The data used for prediction of google stock prices are different for training and testing.
slides are not in sync with speech..but a good explanation
Thanks for the feedback. We will look into it. Cheers!
Any tutorial which shows tuning of LSTM that is build using keras.
Hi Imran,
We don't have any relevant tutorial that shows how you tune LSTM hyperparameters with Keras. Here is a blog that can help you out in this
machinelearningmastery.com/tune-lstm-hyperparameters-keras-time-series-forecasting.
Thank you for an excellent tutorial and also for using a practical real life example. I was going round in circles trying to figure out what was the last parameter as one when we were reshaping in numpy array and now I understand it is just an indicator that the number of parameters is now ending. In this example you have used three layers of lstm ? Is there any information on how the learning process and also the outcome would vary when we used different layers of Lstm
Hi, There are 4 Lstm layers used while training the RNN model. To be fair, there is no proper way to determine a network topology just from the number of inputs and outputs. It depends on the number of training samples and the complexity of the model you are trying to train. In RNN, 4-5 LSTM layers will generally produce a good model. You will find a variation in the output to a very small extent. Hope this answered your question. Please let us know if in case you have any other queries. Thanks for watching the video. Cheers!!
Hello sir
after running the below command
# Getting the prediction stock price of 2017
dataset_total= pd.concat((dataset_train['Open'], dataset_test['Open']), axis=0)
inputs = dataset_total[len(dataset_total) - len(dataset_test) - 60:].values
inputs= inputs.reshape(-1,1)
inputs= sc.transform(inputs)
x_test =[]
for i in range(60, 80):
x_test.append(inputs[i-60:i, 0])
x_test = np.array(x_test)
x_test= np.reshape(x_test, (x_test.shape[0], x_test.shape[1],1))
predicted_stock_price= regressor.predict(x_test)
predicted_stock_price= sc.inverse_transform(predicted_stock_price)
i got this error ### ValueError: Found array with dim 3. None expected
I have used this for weather prediction. The predicted value of training and testing when plotted against the actual value are seems to be shifted by 1 step, can you explain the reason.
Hi Imran,
Here are the steps you can try to solve this issue:
1. Decrease the LSTM dimension
2. Add some form of regularization to combat overfitting (dropout)
3. Train for more epochs or change the learning rate.
@@SimplilearnOfficial I have tried all the steps that you have mentioned. Can you give me some idea how to decide LSTM dimension. I mean from which size to start. For eg. In my case I have 15000 rows and 8 coloumns in training dataset. So from how many units should i start in LSTM layer and also how many number of layers.
Thankyou.
In most of my cases it happens that loss during training keeps on decreasing but validation loss is changing between some fixed range. Any suggestions for this???
Thankyou.
"Hi Imran,
This happens due to improper hyper-parameters used while training. Please adjust the hyper-parameters and train the model for more number of epochs."
"Hi Imran,
Check out this link for choosing the right hyperparameters for LSTM towardsdatascience.com/choosing-the-right-hyperparameters-for-a-simple-lstm-using-keras-f8e9ed76f046"
How many years is the real google 2017 dataset I’ve tried to do SPY dataset but the graph didn’t come the same way it was shown according to the video
Hi Karim, Google_Stock_Price_Test dataset has data only for Jan 2017. The model is trained from the data that has values from 2012 to 2016.
I am getting the error numpy.ndarray' object has no attribute 'append' when I run the command X_train.append(training_set_scaled[i-60:i, 0])...Any idea how to solve this ?
"Hi Devesh,
Please refer to the following link that will help you solve this issue: stackoverflow.com/questions/42942781/python-attributeerror-numpy-ndarray-object-has-no-attribute-append/"
Is there any video regarding use of LSTM for pixel classification?
LSTMs are generally not used for image classification and identification tasks. Convolution Neural Networks work better where inputs are image pixels.
@@SimplilearnOfficial If we consider that neighborhood pixels are similar then a small patch of image can be classified using LSTM.It is implemented in an IEEE paper.
Thanks for your input. It is true that LSTMs can be used for general image classification problems. But not for implementing any major visual imagery task.
@@SimplilearnOfficial can you upload any pixel classification technique tutorial .
thanks for you giving this training free on youtube, i think if you have a github account sharing files used in video, that would be nice.
Hey Oz, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we can send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that too. Hope that helps.
Please let me know how can I download the slides . I could not do it on slideshare for some reason
Hi Mohamed, we have restricted download due to company policy. Thanks.
Tutorial seemed really helpful,
Please provide the code!!!
samriddhi100mit@gmail.com it's my mail ID
Hi, thanks for watching our video. We have sent the requested dataset to your mail ID. Do show your love by subscribing to our channel using this link: ruclips.net/user/Simplilearn and don't forget to hit the like button as well. Cheers!
How reshape function working there
Reshape function is used to reshape the input data. When called on an array, the reshape must evenly reorganize the data in the array.
Can I get the dataset?
Hello Ankush, thanks for watching our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
@@SimplilearnOfficial hi i was wondering if i could get dataset on award2005@gmail.com thank's advance
Hi, thanks for watching our video. We have sent the requested dataset to your mail ID. Like and share our video with your peers and also do not forget to subscribe to our channel for not missing video updates. We will be coming up with more such videos. Cheers!
why dont you just attach the dataset in the description instead of just emailing it to each and every person, like dam
Sorry about that! We have sent the requested dataset to your inbox. Thanks.
Slides are running fast than audio.
Thank you so much for bringing this to our attention. We reported this right away to the relevant department.
"Hate cannot drive out hate; only love can do that." author ~ "Martin Luther King Junior"
Nice quote! Have a good day!
@@SimplilearnOfficial AI is like humanity's child. Best to instill ethics and morals now
Well said. Cheers!
Thank you for the great video! I have a problem however with the real forecast of the model. Looks like it can be only use to compare the already existing data, but is very difficult to make it predict non existing values. Can you also make a video about that part? Should the model be saved and after that used may be?
Hello, thank you for watching our video. We will definitely look into your suggestions. Do subscribe and stay tuned for updates on our channel. Cheers :)
"Karas" and "num-pee"... I honestly couldn't tell whether or not this guy was trolling everyone on the internet... Good video. though
For everyone else: it's "Keras" (you wont find anything related to "Karas"). Additionally, "numpy" is pronounced "num-pie", just like "python" ("pie-thon")... unless you pronounce it as "peethon" and you like to fondle/"Karas" sweaters... :p
great video though... were you trolling the internet? I seriously can't tell.
No, it must have been the accent issue. We definitely share the feedback to the concerned team and make sure it is not happening again. Thanks for your support and love!
Is this code available on github? If not I've DM'd you my email and would be thrilled if you could send it my way. Thanks for the helpful tutorial!
Hello Nick, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.
Great content but cant listen to more than 5 mins of that terrible low quality scratchy tinny sound quality, please reprocess for sound quality. :-((
Thank you for your review. We are sorry to hear you had such a frustrating experience, but we really appreciate you bringing this issue to my attention
use animation .not verbal explain
Hey Aref, thank you for watching our video and for the honest feedback. We will definitely look into this. Do subscribe, like and share to stay connected with us. Cheers :)
Why are you guys doing this tutorial if you can't provide the dataset. I have been waiting for 3 days now... Total waste of time
Hello Ravi, thanks for viewing our tutorial and we hope it is helpful. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly.
Excellent explanation...thanks
Hey Reddy, thank you for appreciating our work. We are glad to have helped. Do check out our other tutorial videos and subscribe to us to stay connected. Cheers :)
Can i get the data set???
Hello, thanks for viewing our tutorial. It would be helpful if you will provide your email ID to us so that we could send the requested dataset promptly. On the off chance that you need your email ID to be kept hidden from others, we can do that also. Hope that helps.