Iris data set

Figure 12.2 Scatter plot of the first two PC scores obtained from PCA analysis of the Fischer iris data set.

Table 8.2 The first ten samples from Fischer s iris data set, used to demonstrate PCA compression...

Table 8.3 The explained variance in the X-data, for each PC, for the iris data set...

The CD that accompanies this book contains a well-studied set of data known as the Iris dataset. A brief description of the data is included on the CD. Write a GCS to analyze the Iris data. If you have SOM software available, compare the performance and execution time of the SOM and the GCS on this dataset. 

Earlier it was mentioned, and demonstrated using the Fisher Iris example (Section 12.2.5), that the PCA scores (T) can be used to assess relationships between samples in a data set. Similarly, the PCA loadings (P) can be used to assess relationships between variables in a data set. For PCA, the first score vector and the first loading vector make up the first principal component (PC), which represents the most dominant source of variability in the original x data. Subsequent pairs of scores and loadings ([score vector 2, loading vector 2], [score vector 3, loading vector 3]...) correspond to the next most dominant sources of variability. 

An example of PCA compression is made using the classic Fisher s Irises data set.29 Table 8.2 lists part of a data set containing four descriptors (X-variables) for each of 150 different iris samples. Note that these iris samples fall into three known classes Setosa, Verginica, and Versicolor. From Table 8.2, it is rather difficult to determine whether the four X-variables can be useful for discriminating between the three known classes. 

A set of IRIS data[ 17] which consists of three classes setosa, versicolor and Virginia was used to determine the applicability of the PP PCA algorithm for analyzing multivariate chemical data. Figure 5 showstheclassificationresultsofPPPCAandSVD.lt can be seen that the PP PCA solutions provide a more distinct separation between the different varieties. 

Measurement of binding constants (Kj) or 50% inhibition concentrations (150) for Iri vitro events usually provides data of far higher precision than that of Iji vivo studies. When carried out on a series of structurally related compounds, the data set (logl/Kj or pI50) can be analyzed in terms of a linear combination of relevant descriptors to provide a mathematical model of the event. The number of descriptors the model will support depends partly on the size and data span of the set and partly on the strength of statistical mea-2... 

Information in the files for the RfD and RfC can be useful for identifying a risk of reproductive and developmental toxicity in humans, if such data have been used in setting the RfD or RfC. The review of data in establishing RfDs and RfCs is comprehensive, and newer assessments have detailed support documents that can be downloaded from the IRIS Internet site, chttp // www.epa.gov/iris>. 

Common Data Source Approach. Another approach is to use a common source of data for benchmarking a set of chemicals. Examples of potential data sources include Material Safety Data Sheets (MSDSs) from product manufacturers, the Hazardous Substances Data Bank (HSDB), the Integrated Risk Information System (IRIS), the International Uniform Chemical Information Database (lUClID), the High Rroduction Volume Information System (HRVIS), the Organisation for Economic Cooperation and Development (OECD) Screening Information Dataset (SIDS), and the Canadian Domestic Substances list... 

In the above form, the Carrean model can be fitted to the entire viscosity versus shear rate curve. However, such a complete set of data up to iri is rarely determinable. Hence, the popular form of the Carrean model that is used as the truncated three-parameter model after neglecting is given below ... 

If data for reactions with all five members of the Basic Monomer Set are not available, use can be made of such data as exists, always provided that styrene and acrylonitrile are among the monomers ineluded. This condition ensures that the data are spread over a wide range of radical polarity, represented by iri, beeause styrene has a very low value (zero), while acrylonitrile has one of the highest values known (0.701). 

---