I understand that all the "Log" scales in TuringBot, both in the Plot Scale on both the Regression and Analysis tabs use the symmetric SymLog() function. I would very much prefer a normal Log10 scale, as I invariably use a Log scale for nonnegative values. One consequence of SymLog is that when plotting Observed vs Predicted, the Plot Scale option "log y and log x" is disconcertingly curved even when the regression is actually perfect (i.e. predicted=observed). This is due to SymLog scaling the x- and y-axes differently.
As a compromise, you could for example give users both options in the Plot Scale dropdown list, i.e. log x, symlog x, log y, symlog y, log x & y, symlog x & y. The log options can simply omit plotting negative values.
