Ceramic Strength

This practical will use a case study from NIST Engineering Statistics Handbook analysing the effects of machining factors on the strength of ceramics to demonstrate visualisation using pandas. And much of the analysis/discussion follows theirs.

Background / Data

Dataset JAHANMI2.DAT

The data for this case study were collected by Said Jahanmir of the NIST Ceramics Division in 1996 in connection with a NIST/industry ceramics consortium for strength optimization of ceramic strength The motivation for studying this data set is to illustrate the analysis of multiple factors from a designed experiment

The response variable is a measure of the strength of the ceramic material (bonded Si nitrate). The complete data set contains the following variables:
- Factor 1 = Observation ID, i.e., run number (1 to 960)
- Factor 2 = Lab (1 to 8)
- Factor 3 = Bar ID within lab (1 to 30)
- Factor 4 = Test number (1 to 4)
- Response Variable = Strength of Ceramic
- Factor 5 = Table speed (2 levels: 0.025 and 0.125)
- Factor 6 = Down feed rate (2 levels: 0.050 and 0.125)
- Factor 7 = Wheel grit size (2 levels: 150 and 80)
- Factor 8 = Direction (2 levels: longitudinal and transverse)
- Factor 9 = Treatment (1 to 16)
- Factor 10 = Set of 15 within lab (2 levels: 1 and 2)
- Factor 11 = Replication (2 levels: 1 and 2)
- Factor 12 = Bar Batch (1 and 2)
The four primary factors of interest are:
- Table speed (X1)
- Down feed rate (X2)
- Wheel grit size (X3)
- Direction (X4)

The goals of this case study are:

Determine which of the four primary factors has the strongest effect on the strength of the ceramic material
Estimate the magnitude of the effects
Determine the optimal settings for the primary factors
Determine if the nuisance factors (lab and batch) have an effect on the ceramic strength

To download the data, run the following code

import os, urllib
if not os.path.exists('JAHANMI2.DAT'): 
    print ('Downloading remote data file ...')
    urllib.request.urlretrieve('https://www.itl.nist.gov/div898/handbook/datasets/JAHANMI2.DAT', 'JAHANMI2.DAT')

The original analyse was carried out in R, we are going to reproduce this in Python/pandas.