Exploratory Data Analysis

On the Analysis tab, in the Exploratory Data Analysis group:

• Click Histogram to calculate and display distribution statistics for a numeric field in a file. You can display histograms, cumulative frequency curves, or probability plots, using actual values or Ln (Natural Log) transformed values.

• Click Histogram (Multivariate) to calculate and display distribution statistics for the numeric fields in a file, and show multiple variables on separate graphs.

• Click Box and Whisker to visually summarise and compare the distribution of values of multiple variables. Any number of variables may be used. They must all be fields in the same file, but there is no need to keep the integrity of individual rows. A Box and Whisker chart includes options for highlighting outliers, showing the drift between individual means, and displaying confidence bands. Confidence bands refer to the confidence of the mean, not the data set.

• Click Scattergrams to calculate correlation statistics for any variables in a file. Both simple linear (when there is only one X-axis field) and multiple linear (when there is more than one X-axis field) regressions can be calculated and displayed.

  Different (exponential, logarithmic, polynomial, power and moving average) regression types are supported whenever there is only one X-field. Optionally, the regression line and its confidence bands may be displayed and colour coded. Confidence bands refer to the confidence of the mean, not the data set.

• Click Q-Q Plot to create a Quantile-Quantile (Q-Q) Chart or a Probability-Probability (P-P) Chart.

• Click Variance-Covariance to display variance-covariance and correlation matrices for multivariate data. To use this function, specify the name of the File and the Field(s) to calculate, specify allowable Minimum and Maximum values, and apply an on-the-fly upper Cut value.

• Click PCA to use Principal Component Analysis to reduce the dimensionality of large multivariate data sets.

• Click Clustering to use a clustering algorithm to identify the data clusters in a file, based upon the values in one or more selected fields.

• Click Dendrogram to load or create a dendrogram for displaying hierarchical clustering of data.

• Click Ternary Diagram to construct a triangular graph, using the data from three field, in which the position of each point is determined by the ratio (percentage) of the three components.

• Click Spider Graph to show the relationships between multiple fields in successive records, and the values for each record as a connected group of points. Each record group is shown in a different colour, associated with the sample ID. Field labels are shown on the X axis and values are plotted on the Y axis. You can interactively view next and previous record sets on the display screen.

• Click Boundary Analysis to assess the contact relationship between two domains.

The boundary analysis works by generating spatial statistics of the domains under investigation, starting at the contact (zero distance) and binning the data at regular distance increments until the maximum distance is reached or there is no more data.