Sure. Use
l(x) = a*x + 100
as model function and fit the parameter a.
xdata = ...; % your xdata as column vector
ydata = ...; % your ydata as column vector
a = xdata\(ydata-100)
But starting from a certain x-value, the slope seems to change. You should separate the fit into two parts:
l1(x) = a*x + 100 0 <= x<= x1
l2(x) = b*x + [(a-b)*x1 + 100] x1 <= x <= x(end)
