Live Day 2- TimeSeries,ETS,EWMA,ARIMA,SARIMAX, Fbprophet Session

[@MohammadrezaMokhtari-qh2yg]

im truly grateful for that smooth and simple teaching! thank you man, thank you

[@rajbir_singh0517]

Hello Sir, you are doing great for the community. I have a request, please explain each and every topic With why it is needed,what can be done and how can be done. This way we can relate with realtime

[@kibetwalter8528]

Hi Krish. Please do an example for the difference between using LSTM for classification and LSTM for regression. Explain the difference between using LSTM for the two. Especially for multivariate. You have always been my teacher. I learned machine learning and deep learning from you. No other bootcamp, I didn't do any computer science course in University. Just your KZbin videos. Thank you so much.

[@TheBarinco]

I think a really valuable video that can be done is for count-based time series modeling. It does not seem to be covered very often...

[@assayebelay6111]

Really, you are wonderful!

[@rahulshelke7624]

This sessions are intresting

[@tagoreji2143]

Thank you, Sir. It was a very good session

[@kumarpriyansh4238]

Hello sir . Nice video ..but AR or MA that you explained is just high level discussion. Like how you gonna find phi matrix and most importantly how do we know past error say e_t-1 , e_t-2 . In MA everything on left side of equation is unkown. Can you make in depth discussion about innovation algorithm or durbin lavison algorithms , log-liklihood, yule walker equations etc... If you can show us how to implement MA or AR from scrath without using inbuilt arima method...then it would be much more helpful. I wanted to know what is going on behind the scene.

[@harshvardhanmutatkar738]

Quick Note guys I know you didn't understand 1:13:59 this part clearly... Moving Average Model is a tool used to predict values in a time series by looking at the noise or unexpected changes from previous time points. Which means in time series each new valve is often influenced by unexpected changes (like errors) from recent points. The MA model looks at a Fixed number of these past errors to make prediction. For Example: Imagine we are trying to predict daily temperature The idea we came up is MA (1) model, in this today's temperature prediction Based on can be adjusted yesterday's error (how much we missed yesterday). μ: In this model we consider μ or mean as average of all the series of their historical data points in this. Model μ as Constant for prediction further value, I don't know why krish has not taken it as constant or he forgot to or may be this be another method. θ: 'tita' it is a coefficient, that determines how much influence the previous error (the difference between the predicted and actual value from previous time step) has on the current prediction. In some context θ can be Constant based on understanding of the data how sensitive you want the model to be on recent errors. Larger θ (close to 1): The model responds more to recent errors, making it more sensitive to changes but potentially more volatile. Smaller θ (close to 0): The model is less responsive to recent errors, leading to smoother predictions but possibly slower adjustments to new trends. θ is crucial in determining how much past errors will influence future predictions in an MA model. The predicted MA by Krish by keeping the μ constant should be: 10,9,10.5,9.5,11. Thank you.

[@abdullahsakib7306]

many many thanks

[@enareshy]

Thank you Krish..

[@gh504]

Crystal clear explanation thank you sir

[@atulkadam6345]

Just tell me how to find error term.. Et-1

[@amitchauhan6831]

Everyone is predicting values on test data till the date we have in dataset, what about the prediction values beyond the last day/time in the dataset.

[@BhavinShah74]

Main concepts ARIMA and all me consistency bigad gaya.. baki initially basic concept was nicely conducted.. thanks..

[@sangitamishra1730]

🙏🙏🙏

[@Saurav_suman-pj8uk]

we all love u sir, just wanna meet you once

[@CTO371]

Sir make more and more videos on statistical like ANOVA, sampling methods like simple simple random sampling, stratified, systematic etc....

[@mbmathematicsacademic7038]

Im enjoying this

[@gauravfamily2209]

how to choose which function to use expanding and rolling for window size? pls guide me.

[@nitinsinghbisht8959]

How come the window size is set to 5 but the addition is for first 6 entries doesn't makes sense? @Krish naik

[@vadimcosman5480]

Very good question. Doesn't make sense for me either.

[@TheInnomind]

By mistake.

[@siddheshdhanawade3709]

how can i get this file , the git link given is for previous session, plz upload these files to the git hub link you provided

[@AkshayKumar-cw4hd]

According to me windows moving avg is in In statistics subject called sqc we seen that quarterly and yearly avg

[@Mani_Ratnam]

For stocks ,we need to design a different model and it might work for certain stock ranges

[@meeram363]

Sir please explain numeric example for solving ma,arma and arima.Those kind of material is not available.Please help sir,its an important question for our exam

[@SANJUKUMARI-vr5nz]

Good evening sir

[@ABHI43218]

jupyter notebook and notes of this lecture is not available in github link and community section of ineuron

[@krishnabhutada3983]

The file of day 2 has not been uploaded....plz do it asap.

[@sakthivell5641]

Hi, the pdr.get_data_yahoo('---') command is not working, Is there any other alternate way to get data from yahoo?

[@alihaiderabdi9939]

Sir also 1 video on time series statistical models like croston, naive bayes, xgboost

[@PardeepKumar-kh9vv]

Kindly upload the jupyter file for this video.

[@prashantpal3653]

in simple moving average the first average is wrong as it contains 6 observations

[@SahilRajput-xe6uj]

sir if formula is this: u+Q1⋅Et−1, then u is constant na but you are taking u as previous value plz reply to this i need clarification

[@khushisharma-tp7ff]

sir deep learning with time series too pleease

[@laizerLL572]

Hi Sir, you are doing great. I have a request, I faced a challenge to make predictions using univariate time series data but the sadness is my data have a white-noise process, please assist me on how to handle the situation

[@muhamadkamran7886]

Sir I had question about autoscraper tutorial about web scrapping "How to we create DataFrame in Auto Scraper & How can we save scraped content in csv file in AutoScraper"

[@PiyushSingh-uz4yc]

Need a video on hierarchical timeseries

[@artialex6289]

How to get ur notes?

[@RajaReivan]

what is lag?

[@sivachaitanya6330]

Krish.........I dont find the day 2 timeseries jupyter notebook file in the github please upload it............

[@vadimcosman5480]

Wooow. I like very much some of your explanations but doing 15+17+11+12+5 "with respect to the calculator", that is quite 🤔. I do understand when you do 68/5 but the simple one

[@joeljoseph26]

EMA = EWMA. both the same.

[@vishnuprakash9196]

Yeah! the equations too. We can rearrange both equations to reach the other one.

[@atifroome]

Let’s find out how many mistakes he did that went unnoticed by him. write down how many you counted

[@atifroome]

I found 10 mistakes

[@VR-fh4im]

Wrong.