Normal distribution, data representation

In contrast to the large variety of averages and measures of dispersion prevalent in the literature, the number of basic distributions which have proved useful is relatively small. In droplet statistics, the best known distributions include the normal, log-normal, Rosin-Rammler, and Nukiyama-Tanasawa distributions. The normal distribution often gives a satisfactory representation where the droplets are produced by condensation, precipitation, or by chemical processes. The log-normal and Nukiyama-Tanasawa distributions often yield adequate descriptions of the drop-size distributions of sprays produced by atomization of liquids in air. The Rosin-Rammler distribution has been successfully applied to size distribution resulting from grinding, and may sometimes be fitted to data that are too skewed to be fitted with a log-normal distribution. 

At this point let us return to the aluminum content data of Table 5.1. The skewed shape that is evident in all of Figs 5.2—5.5 makes a Gaussian distribution inappropriate as a theoretical model for (raw) aluminum content of such PET samples. But as is often the case with right skewed data, considering the logarithms of the original measurement creates a scale where a normal distribution is more plausible as a representation of the phenomenon under... 

Data Representation. Transformations can be applied to the data so that they will more closely follow the normal distribution that is required for certain procedures or for removing (or lessening) unwanted influences. Certainly for data analysis in which major, minor, and trace elemental concentrations are used, some form of scaling is necessary to keep the variables with larger concentrations from having excessive weight in the calculation of many coefficients of similarity. 

The representation of this equation for anything greater than two variates is difficult to visualize, but the bivariate form (m = 2) serves to illustrate the general case. The exponential term in Equation (26) is of the form x Ax and is known as a quadratic form of a matrix product (Appendix A). Although the mathematical details associated with the quadratic form are not important for us here, one important property is that they have a well known geometric interpretation. All quadratic forms that occur in chemometrics and statistical data analysis expand to produce a quadratic smface that is a closed ellipse. Just as the univariate normal distribution appears bell-shaped, so the bivariate normal distribution is elliptical. 

If a parametric distribution (e.g. normal, lognormal, loglogistic) is fit to empirical data, then additional uncertainty can be introduced in the parameters of the fitted distribution. If the selected parametric distribution model is an appropriate representation of the data, then the uncertainty in the parameters of the fitted distribution will be based mainly, if not solely, on random sampling error associated primarily with the sample size and variance of the empirical data. Each parameter of the fitted distribution will have its own sampling distribution. Furthermore, any other statistical parameter of the fitted distribution, such as a particular percentile, will also have a sampling distribution. However, if the selected model is an inappropriate choice for representing the data set, then substantial biases in estimates of some statistics of the distribution, such as upper percentiles, must be considered. 

Figure 1. Connectivities and principle bonding properties of carbon. From top to bottom connectivity, chemical bonding representation, distribution of n electrons, hybridization symbol, bond length, orientation of the n bonds relative to the carbon skeleton. The spectra represent polarization-dependent carbon 1 s XAS data for sp2 and sp3 carbons. The angles denote the orientation of the E vector of the incident light relative to the surface normal of the oriented sample. The assignment of the spectral regions is given and was deduced from the angular dependence of the intensities of each feature. The graphite impurity in the CVD diamond film is less than 0.1 monolayers.

Cartesian coordinates are a convenient alternative representation for a spatial distribution function. Being uniform over the local space, the data structure obtained is easy to represent (access), to normalize, and to visualize. Use of a Cartesian representation becomes a necessity for complex or very flexible molecules. The principal drawbacks of this coordinate system are the size of the data structure it generates (typically about 1,000,000 elements), its inherent inefficiency (since the grid size is determined by the shortest dimension of the smallest feature one hopes to capture), and the fact that its sampling pattern is usually not commensurate with the structures one wants to represent (which can cause artificial surface features or textures when visualized). Obtaining sufficiently well-averaged results in more distant volume elements can be a problem if the examination of more subtle secondary features is desired. See Figures 7, 8 and 9 for examples of SDFs that have utilized Cartesian coordinates. 

The two intensity levels correspond to the two life times of the excited state of the TMR molecule which is due to electron transfer from neighboring Guanosin bases to TMR [27], [31]. We have chosen to calculate the correlation function (Fig. 4.8c) which shows the chemical relaxation rate of the conformational transition. The correlation function cannot be fitted with a single exponential as would be normal for first order kinetics. Instead, a stretched exponential of type exp-(kt) gave the best representation of the measured data. The stretched exponential can be envisaged as a distribution of relaxation rates with a mean relaxation rate k and a distribution defined by the... 

The classification methods discussed in the previous section are all based on statistical tests wliich require normal data distribution. If this condition is not fulfilled the so-called non-probabihstic , non-parametric or heuristic classification techniques must be used. These techniques are also frequently referred to as pattern recognition methods. They are based on geometrical and not on statistical considerations, starting from a representation of the compounds... 

---