Home 
Search 
Today's Posts 
#1




solver can not solve polynomial eq. where one variable is five di.
I tried to create equations from the set of data by solver
x= 90021 85248 78314 69768 59508 47975 y = 20.09 19.92 18.70 16.95 16.13 15.67 in the form of y= a + bx +cx2 + dx3; where x2= x square ,x3 = x cube. but could not do so. the solver does fine when the value of x is small . Your help will be hghly appreciated in this regard. Thanks 
#2




You might want to check out solver.com to find out exactly what the
limitations and weak points for the buitlin Solver are. tj "Amitava" wrote: I tried to create equations from the set of data by solver x= 90021 85248 78314 69768 59508 47975 y = 20.09 19.92 18.70 16.95 16.13 15.67 in the form of y= a + bx +cx2 + dx3; where x2= x square ,x3 = x cube. but could not do so. the solver does fine when the value of x is small . Your help will be hghly appreciated in this regard. Thanks 
#3




"Amitava" wrote...
I tried to create equations from the set of data by solver x= 90021 85248 78314 69768 59508 47975 y = 20.09 19.92 18.70 16.95 16.13 15.67 in the form of y= a + bx +cx2 + dx3; where x2= x square ,x3 = x cube. but could not do so. the solver does fine when the value of x is small . Your help will be hghly appreciated in this regard. In Excel you're limited to 15 decimal digit precision most of the time. A few intermediate calculations may have more if they can stay in FPU registers. However, standard minimizing sum squared error fitting would at some point involve squaring the x^3 values, at which point you've blown way past machine precision. That said, plot these XY values in an XY chart, add a polynomial trend line or order 3, and display the trend line equation. It'll show the a, b, c and d coefficients you seek. 
#4




You can use LINEST to find equation of best fit.
See www.stfx.ca/people/bliengme/ExcelTips  Bernard Liengme www.stfx.ca/people/bliengme remove CAPS in email address "Amitava" wrote in message ... I tried to create equations from the set of data by solver x= 90021 85248 78314 69768 59508 47975 y = 20.09 19.92 18.70 16.95 16.13 15.67 in the form of y= a + bx +cx2 + dx3; where x2= x square ,x3 = x cube. but could not do so. the solver does fine when the value of x is small . Your help will be hghly appreciated in this regard. Thanks 
#5




I imagine you could tweak Solver's parameters and/or scale your own
values to get Solver to work, but, in this specific instance, you should consider using LINEST as Bernard suggested. If you did, you would discover that none of the coefficients for an order 3 fit are statistically significant at the 5% level. Only the constant term is significant for a quadratic fit whereas both terms are for a linear fit. To learn more about how to test for statistical significance check the LINEST help Example 5.  Regards, Tushar Mehta www.tusharmehta.com Excel, PowerPoint, and VBA addins, tutorials Custom MS Office productivity solutions In article , says... I tried to create equations from the set of data by solver x= 90021 85248 78314 69768 59508 47975 y = 20.09 19.92 18.70 16.95 16.13 15.67 in the form of y= a + bx +cx2 + dx3; where x2= x square ,x3 = x cube. but could not do so. the solver does fine when the value of x is small . Your help will be hghly appreciated in this regard. Thanks 
Reply 
Thread Tools  Search this Thread 
Display Modes  

