2. Explain in detail:
SST
55k
SSE
Ans) SST
The sum ofsquares tal, denoted $57, is the squared differences
between the observed dependentvariable and it mean. You can tink
of this as the dispersion o f the observed variables around the mean —
much like the variance in descriptive statsics.
The Total SST tells you how much voriafon here isin the dependent
voriable. Total $9 = (Yi - mean ofY) 2 find the acval number that
represents asum ofsqvares.A diagram (like he regression line above)
is optonal and can supply a visual representaton ofwhatyoure
calcvlatng.
Stis a measure of the total variability of the datset.
SSR
The second term is the sum ofsquares due regression, or SSR. Ht
is the sum ofthe differences bet cen the predicted valve and he
mean of the dependentvariable. Think ofitas ameasure thatdescribes
how well ourline i the data,
Bis value of 598 is equal » the sum ofsquares tal, itmeans our
regression model cop-wres all the observed variability and is perfect.
O nce again, We have mention tatano ther common notion is ESS
or explained sum of squares
S52 = 4( 4) 22 SST - SSE. Regression sum of sqvares is interpreted
aseerrorsum of squares is obtained by firstcom puting +he mean
aeofeach battery 4 pe. Foreach battery ofa specified ype, he
is subtracted from each individual battery's lifetime and then
ed. The sum of these squared terms forall battery types equals
. SSE is ameasvre of sampling error.
lastierm is the sum of squares error 0 r SSE. The erroris the
ence between he observed valve and the predicted valve.
sually Want minimize te error. The smaller the error, the better
timation power of the regression. Fimolly, Lshould add hatitis
now n as residval sum ofsquares.