is it possible to fit a quartic regression 4rth order with arduino ?
y = e-dx+cx^2 +bx^3+ax^4
y=3.99915 - 0.03974209x + 0.0001509082x^2 - 0.0000002300598x^3 + 0.0000000001262231x^4
Of course it is...
Regards,
Ray L.
You may have trouble preserving this level of precision (0.0000000001262231) with an Arduino float.
Why do you want to fit a curve using Arduino? Give us the big picture.
Many arduinos (e.g. Uno) only have six digit precision in floats.
I Made a current sensor using an LM358 and a SHUNT resistor for my back up power supply.
following are the values that i received after loading .
ADC Ampere
224 0.4
412 0.8
535 1.0
607 1.2
672 1.4
729 1.6
734 1.8
745 2.0
am trying to find an equation to fit the ADC>Ampere
plotted those values in excel and looks to me a 3rd order curve fits nicely with coefficients within the arduino float realm! 
alternatively...
why not just create a lookup table for current vs ADC and use that on your arduino?
@sibin_lal, what are you using for a voltage reference? (Please say it is not USB power.)
@Coding Badly i use a Arduino NANO which is powered by an external 9V linear regulator . i guess the voltage reference still is 5v . i do not have any external reference.
@sherzaad thank you 3rd degree would do but I still don't have a fair algorithm. 
sibin_lal:
following are the values that i received after loading .
Are those readings reproducible?
@Coding Badly yes they evidently seems to be reproducing the values the curve is stable
sibin_lal:
@sherzaad thank you
something like this you mean?
void setup() {
uint16_t adc_read;
float calc_current;
Serial.begin(115200);
analogRead(A0);
adc_read = analogRead(A0);
//calc current (mA) = (0.6715*adc_read)-(0.0001*(adc_read^2))
calc_current = (0.6715 - (0.0001 * adc_read)) * adc_read;
Serial.print("calc current (mA): ");
Serial.println(calc_current);
}
void loop() {
// put your main code here, to run repeatedly:
}
I've replotted the value for ADC vs I this time and to me a quadratic seems to do the job!
Blackfin:
You may have trouble preserving this level of precision (0.0000000001262231) with an Arduino float.
= 1.262231 x10-10
Looks fine to me. 7 digits of precision. Maybe the last digit might get mangled by a single-precision float.
sherzaad:
I've replotted the value for ADC vs I this time and to me a quadratic seems to do the job!
No residuals?
I Made a current sensor using an LM358 and a SHUNT resistor
If that is at all nonlinear, something is wrong. Your amplifier design could be wrong, the resistor is heating up, etc.
Post a schematic diagram with power supply details, component values and ratings (hand drawn, not Fritzing). Note that the LM358 is not a rail-to-rail amplifier and the common mode input voltage may not exceed Vcc-2 V.
sherzaad:
something like this you mean?void setup() {
uint16_t adc_read;
float calc_current;
Serial.begin(115200);
analogRead(A0);
adc_read = analogRead(A0);
//calc current (mA) = (0.6715adc_read)-(0.0001(adc_read^2))
calc_current = (0.6715 - (0.0001 * adc_read)) * adc_read;
Serial.print("calc current (mA): ");
Serial.println(calc_current);
}
void loop() {
// put your main code here, to run repeatedly:
}
I've replotted the value for ADC vs I this time and to me a quadratic seems to do the job! :) 
Thank you sherzaad sorry for the late replay .
with the code you suggested am getting non-variable values
1.5 to 2 amps = 2.01
0.6 to 1.5 amps =1.34
and
less than 0.6 =0.67
i am thinking about removing the (inaccurate) load current value from display and go for an over current disconnect .
jremington:
If that is at all nonlinear, something is wrong. Your amplifier design could be wrong, the resistor is heating up, etc.Post a schematic diagram with power supply details, component values and ratings (hand drawn, not Fritzing). Note that the LM358 is not a rail-to-rail amplifier and the common mode input voltage may not exceed Vcc-2 V.
probably its the circuit .
The circuits is not good. Explanation here
need a small reference voltage to pedestal the output to a known value greater than the amplifier’s minimum output swing.
I'd add, that small reference required to shift up both inputs above specified Input offset voltage of the OPA (LM358).