/sci/ - Science & Math

[Anonymous No.6038183]

hey guys I got a quick question regarding excel.

I'm trying to use the built-in trendline feature to find the best value of 'a' for my model form y=1/(x+a) in a least-squares sense for the following set of data:

x: 5, 10, 15, 20, 25, 30, 35, 40, 45, 50

y: 0.077, 0.055, 0.044, 0.033, 0.03, 0.027, 0.022, 0.021, 0.019, 0.016

the trend seems logarithmic to be but I'm ideally trying to fit that function to the data. Any suggestions?

thanks in advance guys

[Anonymous No.6038187]

any halp?

[Anonymous No.6038213]

bump

[Anonymous No.6038393]

>>6038183
y=1/(x+a)
a=1/y-x

Call z_i the numbers 1/(y_i) - x_i

just compute the average of z_i to get a good approx of a.

You can also use the pre-made function to approximate
z_i with a linear function.
Maybe it will permit you to get a better model.

[Anonymous No.6038426]

>>6038393
I complete my answer :
better have a look here :
http://www.wolframalpha.com/input/?i=+fit+0.077%2C+0.055%2C+0.044%2C+0.033%2C+0.03%2C+0.027%2C+0.022%2C+0.021%2C+0.019%2C+0.016

and make your choice :)
(careful x=your_x/5)