Custom equation related in matlab's curve fitting tool

custom equation : z=a*x+b*log10(y+c)+d (x=M , y=D, z=ratio1)

Regression analysis was performed from the above estimation equation with the data of x,y,z.

Data is attached.

However, as in the graph, the regression analysis result was calculated to be small at x=6.4.

That is, the graph is located below the corresponding value.

How to solve this..... :(