Matlab Question

What - as in 2/7 of (90-65) plus 65 equals 72¹/₇?

First, you have to make an assumption about the data you do have. That assumption is that the data points are not a significant anomaly from a trend.

Second, one must make a subjective judgment what trend these data points lie upon; linear, periodic, exponential, polynomial, etc.

Third, fit a function using one of the curve fitting functions in Matlab appropriate for that curve.

Fourth, insert day 9 into the fitted function.

That looks like an exponential decay curve with a first-order time constant of about 4-5 days.

It might be a fifth-order polynomial affected by a random hypergeometric probability distribution.

{I love to make things needlessly complicated, sometimes.}

Yes. Er, um, what?

Decay? Decay implies a negative slope.

"1 minus (e to the minus k t)", then, not to be negative about it...

Correction: 6¹/₄ days.

I think what you are after is called interpolation, you want the value for December 9.

The function you want is interp1.

Let x be date: x=[1, 3, 7, 14, 28];

Let y be value: y=[16, 40, 65, 90, 99];

x1 = 9;

y1 = interp1(x,y,x1);

There are various options for interpolation.

Here is a help page:

http://matlab.izmiran.ru/help/techdoc/ref/interp1.html

73 is perhaps the closest integer estimate, maybe 74. PW's linear interpolation might not be a bad start, but the general curve indicates a bit more.