cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Showing results for 
Search instead for 
Did you mean: 

Community Tip - Learn all about PTC Community Badges. Engage with PTC and see how many you can earn! X

Mismatch calculation result

VladimirN
24-Ruby II

Mismatch calculation result

Hi,

The Mathcad document have some calculations from the old article. The article says that all the calculations are done on a computer. However, my results do not coincide with those shown in the article. On the screenshots you can find input information (a, t), the formula for calculating the value (x, y, b_0, b_1). I have tried to carry out calculations with different accuracy, but the result is still different. Maybe I missed something?

Pic_1.png

Pic_2.pngPic_3.png

Pic_4_result.png

5 REPLIES 5
Werner_E
25-Diamond I
(To:VladimirN)

I get the very same results as shown in your sheet and the results are pretty close to those given in the handwritten screenshots of yours. Difference for b_1 is just 7*10^-4, for example.

The reason for that small discrepancy could be that the values of a are given rounded to 2 decimals but the original calculations may have been done with more accurate values. So we can't duplicate the calculation exactly without knowing the accurate values of a. You may also notice that even the values of x and y are somewhat different at the second decimal place which is also a hint that the original calculation was not done with the given, rounded values of a.

1.png

EDIT: You may also check using Mathcad's built-in functions:

2.png

Thank you. Too frequent rounding calculations lead to such difference results.

RichardJ
19-Tanzanite
(To:VladimirN)

I have checked this every way I can think of, and I can't figure out how the author got that number. I even tried using the symbolic processor to restrict the numerical precision of the calculations to a few decimal places, but the answer is never greater than 1.

As an aside, you can use the mean function, as Werner shows, to make the expression rather more compact. Going one more step, the expression is the covariance of x and y divided by the variance of x, so you could write cvar(x,y)/var(x). But of course, as Werner also points out, that's just the slope of a linear regression, so you could just write slope(x,y)

Thanks. Yes, I know about the possibility of using the built-in functions here, but I need to show this algorithm for a person who is not familiar with Mathcad. I think the main problem here is the calculation with too frequent rounding of intermediate calculations.

RichardJ
19-Tanzanite
(To:VladimirN)

I think the main problem here is the calculation with too frequent rounding of intermediate calculations.

Maybe, but the float keyword does exactly that, and did not yield the same results. Maybe the author rounded in a different way though.

Announcements

Top Tags