3.6 Taking Uncertainty into Account

<< Click to Display Table of Contents >>

Navigation:  3. Estimating Weight and CoG >

3.6 Taking Uncertainty into Account

Estimating Weight and CoG

In the estimation area, the right-hand column contains information about the standard deviation. The standard deviation can be given either as an absolute value or as a percentage. To switch between these ways of showing the uncertainty, select the Std. Dev. in the Estimation Window.

The deviation of the selected estimation coefficient is calculated from the spread of the coefficients around the regression line. If the number of coefficients is less than three, the deviation cannot be calculated. Then a value based upon subjective judgment can be manually entered into the grid table.

If some of the other parameters are uncertain, an uncertainty value should be entered into the grid table to the right parameter value. When all parameters have got appropriate uncertainty value attached, clicking the Calc.std.dev button in the Estimation area gives the total uncertainty for the weight in this weight group, and this value is stored.

By the use of successive calculation calculation the uncertainty for a weight group containing subgroups, will be calculated on the basis of the uncertainty of the subgroups.

Even though the uncertainty is an approximate method to obtain the uncertainty for a weight and even if you have to set the values yourself for most weight groups, there are several reasons to take the uncertainty into account. First of all you will be able to use your resources where they are most needed. When selecting Most uncertain under the Std. Dev. menu in the main window, ShipWeight will take you to the weight group where a further estimation and splitting of the group will give the greatest impact on the total result. This will help you to focus on the weight groups that really matter at any stage of your estimation. Secondly, taking uncertainty into account will give you a basis to compare the uncertainty between different projects.