Trendline equation

Hi experts, I have some data in ranges C2:C7, and B2:B7. Values are:

for range B2:B7 we have this:
B2: 100
B3: 200
B4: 300
B5: 400
B6: 500
B7: 600

and for range C2:C7 we have this:
C2: 100.41
C3: 200.842
C4: 301.274
C5: 401.706
C6: 502.138
C7: 602.57

and resulting Trend line equation in the graph is the following:

y = 6.50521E-19x3 - 6.39246E-16x2 + 1.00432E+00x - 2.20000E-02
with R2 = 1.00000E+00

I usually obtain directly the coefficients (6.50521E-19, -6.39246E-16, 1.00432E+00, -2.20000E-02) using something like this:

=INDEX(LINEST(C2:C7,B2:B7^{1,2,3}),1)
=INDEX(LINEST(C2:C7,B2:B7^{1,2,3}),2)
=INDEX(LINEST(C2:C7,B2:B7^{1,2,3}),3)
=INDEX(LINEST(C2:C7,B2:B7^{1,2,3}),4)

(each value in a different cell)

but I realized that numbers with so small power (E-16, E-19 for this case) the result coefficient is zero using the formula.

Is there some way to see the correct small coefficient numbers (6.50521E-19 and -6.39246E-16 for this case), just like the graph shows?

Thanks in advance.

Trendlines and LINEST compute coefficients in different ways, for some reason that escapes me, so I don't have a suggestion for that.

However, there is no need to invoke LINEST (a very heavyweight function) four times to get four coefficients. Select a range 4 columns wide, and paste this in the formula bar:

=LINEST(C2:C7, B2:B7^{1,2,3})

Don't press Enter; instead, press and hold the Ctrl and Shift keys, then ppress Enter.

Not really, bstoyano. I only want to be able to see a very small number (that it is practically zero, but not zero, per example: 6.50521E-19) using the formula but I don't want just a simple zero instead.

Thanks, shg. That is what I thought about it. Is there some other way to get the coefficients without using LINEST function and be able to show even the smallest numbers?

You could do your own matrix inversion, but why bother? The numbers are negligably different, considering the dynamic range of the coefficients, and the fact that you're doing a basis-function (here, polynomial) fit, which is, by definition, an approximation of an over-defined system.

Mmmh...Ok, shg. Let me see then what to do on the matter. Thanks a lot.