Recitation for week #11 (starting Sun, Apr 10) and
Assignment #11: Logistic Regression
The problems in this lab should be handed in as homework #11 and are due at noon
on Wed, April 20, 2011.
In this lab, we continue the analysis of the casting problem. Recall that, as part of a longer
manufacturing process, a factory needs to cast a large number of rectangular metallic blocks.
The blocks are manufactured using a mold, consisting of the main cavity, a cup through
which the molten metal is poured, and two risers for cooling. The size and shape of the
pouring cup and risers aﬀect how quickly the metal can be poured into the mold, how
quickly it cools, and whether it sets correctly.
The factory needs to cast batches of 100 blocks of size 4
The current casting approach is fairly conservative – it takes a long time to pour, but
the blocks always set correctly and are usable. Your goal is to achieve a signiﬁcant
reduction in average casting time while still ensuring that most blocks are usable.
Variables characterizing the mold:
The following nine mold-variables can be varied:
Riser Height, Riser Diameter, Riser 1 Position, Riser 2 Position, Gate Diameter,
Cup Height, Sprue Height, Sprue Diameter Bottom, Sprue Diameter Top
To obtain data on how various mold-variable settings aﬀect pouring and cooling, a
batch of 100 is poured with random variations
in the nine mold-variables, centered
about their baseline values. The data are available in
. As before,
the ﬁrst nine columns are the mold-variables listed above and the 10th column is
and the 11th column is
is 1 if the casting is feasible
and 0 otherwise.
This lab project will guide you through the analysis. First, enter the data in the ﬁle
cast = read.csv("castdata2011.csv",header=TRUE)
There are more eﬃcient ways of designing experiments, but we will use this simple random data collection
for now. Design of experiments will be covered later in the course