TI-36X Pro: Stat: x’ and y’ estimates for Exp, Log, Power curves.

[Matt Agajanian]

Hi all.

I see that I can calculate x-estimate and y-estimate for linear/ax+b data right from the STAT mode. I don’t see that function when I use other curve models. Or are there x' and y' functions in stat mode for the other regression models? Although the formula can be stored in f(x) and use num-solve to calculate x’ and y’.

Do a and b carry over into register a & b so I can use them in num-solv for solving the equation placed in f(x) for x’ and y’?

[ijabbott]

(12-04-2022 11:45 PM)Matt Agajanian Wrote:  Hi all.

I see that I can calculate x-estimate and y-estimate for linear/ax+b data right from the STAT mode. I don’t see that function when I use other curve models. Or are there x' and y' functions in stat mode for the other regression models? Although the formula can be stored in f(x) and use num-solve to calculate x’ and y’.

Do a and b carry over into register a & b so I can use them in num-solv for solving the equation placed in f(x) for x’ and y’?

a and b do carry over into variables and the regression formula can be saved to f(x) (or alternatively to g(x) on the TI-30X Pro MathPrint).

Yes, you can use num-solv on the f(x) (or g(x)) function saved by the regression calculation to solve x' for f(x')=y' or (g(x')=y'). Alternatively, a and b variables can be used to calculate x' from the inverse function. For example, for ExpReg ab^x, calculate x' = ln(y'/a)/ln(b) directly on the Home screen.

For y', just evaluate the function f(x') (or g(x')).

As you guessed, the x'() and y'() functions on the 2-Var Stats menu assume linear regression.

[Matt Agajanian]

(12-06-2022 12:08 AM)ijabbott Wrote:  

(12-04-2022 11:45 PM)Matt Agajanian Wrote:  Hi all.

I see that I can calculate x-estimate and y-estimate for linear/ax+b data right from the STAT mode. I don’t see that function when I use other curve models. Or are there x' and y' functions in stat mode for the other regression models? Although the formula can be stored in f(x) and use num-solve to calculate x’ and y’.

Do a and b carry over into register a & b so I can use them in num-solv for solving the equation placed in f(x) for x’ and y’?

a and b do carry over into variables and the regression formula can be saved to f(x) (or alternatively to g(x) on the TI-30X Pro MathPrint).

Yes, you can use num-solv on the f(x) (or g(x)) function saved by the regression calculation to solve x' for f(x')=y' or (g(x')=y'). Alternatively, a and b variables can be used to calculate x' from the inverse function. For example, for ExpReg ab^x, calculate x' = ln(y'/a)/ln(b) directly on the Home screen.

For y', just evaluate the function f(x') (or g(x')).

As you guessed, the x'() and y'() functions on the 2-Var Stats menu assume linear regression.

Yeah. I noticed that the LinReg model is what the 36X Pro uses. Can’t wait for my 30X Pro MathPrint so that the big three (Log, Pwr, Euler Exponential) are available to be calculated.

Then, as you said that the Linear regression model is used for x’ and y’ calculations on the 36X Pro. With this in mind, since Euler’s Exponential model is lacking on the 36X Pro, I just went back to the HP-34C’s Curve Fitting principles to enter data and calculate a, b, r, r^2, y’, and x’.