February 27, 2008
Another set of data has been evaluated. The alum dose for this particular experiment was 40mg/L and the clay dose turbidity was set at 125NTU.
The sum squared error has been evaluated for the n th degree of polynomial regression.
Polynomial regression was evaluated on the raw data. The logarithmic regression was done for comparison purposes. However the equation for logarithmic regression was nonlinear.
Polynomial degree of 3 (n=3) was chosen in this particular set of data and the slope of the curve is plotted below.
Problems that should be looked at
- The initial turbidity is not the same as the clay dose turbidity (Experimental setup)
- Different type of slope plots for different degree, n of polynomial (Plots not shown) - not consistent
- Logarithmic regression is nonlinear
- The slope curve will be smoother n-1 th polynomial if plotted against smaller uniform ¿x
Next analysis
- Saturation plot analysis - After this, will decide which regression line should we choose.
- Draw a smoother slope curve - using smaller uniform ¿x
- Get feedbacks from the Tube Floc group - We Rock!
- Tidy up the programming and make a list of constants used