Regression #4: ANOVA table in regression

[aleatoriu5]

Thank you man, you saved my semester! I learn everything I needed for my final exam in one week just watching tons of your videos about it! Really thanks!

[milklollipop]

this is soo good, helps me in college for one vital course :p thankss!

[suryabhai12]

i am confuse ,i think there is in error in ANOVA table terminology,RSS i think is regression sum of square(actualY-avge Y)is shown as ESS,and other wise,i have refer to last video in which RSS is actually counted which was 51.9,which is ESS ie actualY-fitted Y residual sum of square

my question is RSS is Regression Sum of Square or Residual Sum of Square

[bionicturtledotcom]

@suryabhai12 The sum of squares terminology varies, two basic sets. Above looks fine to me. The sum of the squared residuals is typically called RSS (residual sum of squares) or SSR (sum of squared residuals). The "regression" sums (i.e., sum of predicted Y - average Y) is typically ESS (explained sum). Thanks, David

[ngopalakrishna]

Thank you David!

[zinnatun1644]

Thanks a lot . It was really helpful;

[nayanaran]

the two youtuber who dislike the video must have failed their elementary math...

[nayanaran]

David Harper and Andrew Holmes (CFA Schweser videos) rock!

[musicofthesoul85]

excellent video!

[sureshatt]

Hii,, i'm really thankful for you. but I have a confusion,

most of the books denotes SSR as Regression Sum of Square. and ESS as Error (or Residual) Sum of Square. but you have used it in the other way. can you please explain this.

[nagendrajayanty]

Thanks a lot.. a very good tutorial

[danvedman]

Hi I just want to ask. I'm not sure if it's the right test or not but say I have 3 independent samples and I want to see if there's a signifcant difference between their means or not (one sample is a control sample). What test would I conduct?

[neversummer330]

@danvedman a one-way ANOVA would be appropriate

[ravpothead]

top man 

[hannalindberg]

How do you calculate the p-value?

[neversummer330]

@hannalindberg the practical way is to simply look it up in the F distribution table using the appropriate degrees of freedom

[Zenelope]

graphically:

ESS: distance from green to red

RSS: distance from blue to green

[olimbay]

thanks for your efforts, it was very helpful.

[fallingsnowieflakes]

I still do not find very clear, graphicaly which is the RSS and which the ESS?

Is the RSS the difference between the observation and the average and the ESS the difference between the the regression and the average?

[dpakji]

Good one...

[xxlelexx11]

God bless your soul

[clm211]

thanks for this video. helps clarify

[trknigatu]

Exccellent tutorial. thank you indeed.

Tariku, Ethiopia

[buildstrength]

Hi trknigatu,

need one small clarification.

So, what is regression sum of squares? is it the sum of the square of the distances between regression line and the average line(i.e. red line)

[BrokenHourglass7]

@buildstrength That's right. The sum of the square of the distances between The regression output and the mean (red line). 

[aabbhfg]

ohhh stats is boring

[taityfella]

thanks for your time in making this video it has helped clear up some issues of mine

[juvenileprince]

thanks harper. can i just ask how do you know extent of the strength of relationship between teh independent and dependent variable. Ur calculations show that the coefficient of determination is 0.86 which means the line u have fitted is explaining 86% of the variation in the model? that's gr8 but what's extent of the strength of relationship? how do you find it? using Adjusted R Square?

I majored in Statistics

[Rhizzae]

Thank you very much... this video really helped...

[kumaisobele]

Thank you Sir, clearly explained, extremly helpful

K-from Zambia

[OLDILAENDER]

Wow, that was great! Thank you!

[porthoshsc]

Thanks for this vid. At 7:30 you say we reject the null hypothesis because F is 52.7 which is 'quite high'. What is this to be compared to to decide if it is relevant or high enough? What is the range? Would 2 have been low enough to accept the null hypothesis and 1000 too high?!

[xoxo4440]

the old-school way would be to previously choose a critical F number from the F tables and compare that to the F ratio that you get from your data. if that ratio's bigger than the critical F, then you reject the null. but nowadays PCs simplfy everything w/ their magic p-values!

[binarymax]

Excellent videos, very very helpful.

Regards from London.

[lpswamp]

Very helpful.Regards from Beijing.

[rhettintaipei]

Very Helpful!

Regards from Taiwan

[gogor99]

Thanks a lot Mr. Harper

Is was really very helpful.

Regards from Azerbaijan

[ssukhon]

Great job