Murphy Vista

hello, where i can find your dataset?? can you share?

Dr. Esmalifalak, thank you so much for your teaching in a short time. I wonder that there is any relationship between number of steps and lags in LSTM? What will happens if number of lags variables are far less than number of steps? Technically, it is possible, because the lstm looks for finding coefficients for a function makes between inputs and outputs. Is there any rule to limit number of time steps based on lag variables? for example, if the lag variables are 8, then number of time step (future step for prediction) must be less than 8. Thank you so much for your consideration in advance.

How to download

Hello dr. Mo

I want to hire you as my CDEGS tutor asap. Please reponse

I have very important questions regarding the CDEGS. Please reply if you are existed.

Dr. Moe, I have been looking for you regarding CDEGS If you received this message please reply.

Thank you Dr.Esmalifalak, I have a question regarding the sliding window approach that you used for time series data. Due to the sequence of output, there will be multiple predictions for a single timestep, resulting in overlapping predictions. I am curious about how you handled this while evaluation model ? And plotting the predictions result ? Thanks a lot !!

thanks you profesor, just a question. Is possible deal with categorical variables? Is important the type of enconding to use (one hot or label enconding)? Thanks you in advance

Step stone of DS projects ... Plz make video on it to work with this step with customisable pipelines for different usecases .

Thanks, really helpful video

Thank you for the video. I found it very informative Can you please show how to run .py files for example where do we need to give filepath name and filter city name and can you also please show how the results looks like that are generated from .py file Thank you!

I have a question, this is an unsupervised model, right? is there a way to make the model predict a user input?

Great video!

😘 ᎮᏒᎧᎷᎧᏕᎷ

Thanks for the nice topic. I am wondering if we can do this considering the effect of seasonality? Like, lagging the sales values multiple times and creating new features and then training and testing the anomaly detector?

Excellent Dr.Mohammad. Are these types of algorithms (KNN) considered as weak algorithms in ensemble learning? Please make the similar video and post for other algorithms.

Thanks Dr. Esmalifalak. Your explanation is very useful. How does the accuracy of the program change by changing the step size and log? Will the changes be noticeable? Also, I would appreciate it if you could post a similar video about multi-variable.

Thank you so much. This was really helpful👌

Thanks Dr Mo.

The greatest ml videos in KZbin

Timecodes 0:00 - Intro 0:19 - Problem Definition 2:14 - Importing Data 4:46 - Changing data types - to_datetime 5:48 - Changing data types - LabelEncoder 8:28 - Reindexing - set_index 9:47 - Converting time series to conventional ML problem by shifting dataframe 18:55 - Model training 23:28 - Model evaluation 28:00 - Creating python files for MVP 29:32 - train.py 36:51 - predict.py

Hi Dr. Esmalifalak, I'm a huge fan of all your videos, they've helped me with getting through university and get a career, can you please upload more videos, what data visualization tool do you use?

You can help me with a master’s thesis for my software part (coding) in Python?

You can help me with a master’s thesis for my software part (coding) in Python?

thank sir

can you please provide the code.

Just amazing

excellent!!

Thanks mate.

Sir where is part 2 of this section

It is hard to find such good explanations on Isolation Forest. Keep up the good work!

very helpful video; I want to ask one question about time series part; you have entered n_neighbours=5 why 5? What about if it is 2 or 3 or 4? If I use time series anomaly detection part for 4 - 5 sensors column data; what should I choose for n_neighbours parameter? again 5?

A very well-structured but simple way of explanation. Can we also have a look at measuring the efficacy of the model?

thank you sir.

Please make another video on, Anomaly detection One-class SVM for Novelty detection

Thx, I will apply it~~

Hi, good job, I have a question, how we can resample according to the year?

good video. suggest to turn up the volume. good content nonetheless. thanks

Hi, thanks I found it really helpful, but I have a question about the Contamination parameter, how we can choose a suitable value for this parameter?

I really love your video. could i ask if there is part 2 of 2 for this section? thank you very much!

thanks for this video! its not easy to find high quality content like this! keep it up!

Thanks for the clarification ! after applying iforest , how can I evaluate the cluster's result ? do you have specific method used for evaluation this type of unsupervised learning? I'd really appreciate that.

Hi! Thanks for the great tutorial. But I have a question, is it possible that isolation forest output different result? I have used isolation forest on my dataset, but the output results are a bit different than previous results everytime (I haven't changed any parameter in the model and the dataset I used is the same).

Hello Dr. Mohammad, Is the algorithm effective with the real time streaming data? I have sensor data of around more than 100 sensors, should I need to find the important variables before feeding into the model or should I pass all the variables and let the algorithm decide by itself? Multicollinearity exist in the data .

You solved a big problem for me，thank you

Thank you Mohammad

Great explanation...

Hello! Just a question. Is this an algorithm a classic isolation forest or an extended isolation forest (I saw you named the object with the predictions eif)? Is there any way to implement an extended isolation forest? Basically the difference between EIF and IF is that the EIF takes random intercept and slope and does the split based on the trend line. Thank you for the video!