QuickTour   |   Features List   |   Download   |   Purchase

With CELLULA-P, you can visualise the underlying patterns in large, multi-dimensional datasets and reduce the complexity in the data to a few, relevant descriptors. It can help you to find patterns, identify groupings/similarities and separations/dissimilarities - and to examine their properties.

CELLULA-P provides you options to scale, smooth, or otherwise transform your data prior to analysis. You can also check the validity of your PCA models and determine the degree to which they explain the observed variation.

CELLULA-P Quick Tour

1. Format and Select your Data within Excel

Simply ensure that your data is in Excel in one contiguous block, with the grouping factors and measurement variables in columns; and with your cases, samples, or observations in rows.

Then select CELLULA>Multivariate>PCA from the CELLULA menu on your Excel menu bar - and complete the dialogue box. For a straightforward analysis, using CELLULA-P's default options, that's all you need to do before hitting OK to create the CELLULA-P results page.

2. (Option to) Specify any pre-processing and analysis options

Alternatively, you can use CELLULA-P pre-processing options to mean centre, scale or log transform the columns of data. You can also pre-process rows of data so as to increase the signal-to-noise ratio. You have options, where information about which groups the observations belong to is known, to perform supervised pattern recognition. You can also control the number of components displayed, or use an eigenvalue threshold to determine the ideal number of principal components. And you have control in the estimation of missing data.

3. Review and present your results

CELLULA-P provides a Model Summary: the eigenvalues, the fraction of variation explained by each component (individually and cumulatively), as well as the cumulative fraction of variation that can be predicted by each additional component are provided both numerically and graphically. CELLULA-P tests the significance of each additional component to determine the ideal number of components according to cross-validation.

CELLULA-P provides Scores & Loadings Plots: the scores plot displays the patterns, groupings (similarities) and separations (dissimilarities) of the observations, while the corresponding loadings plot indicates which descriptors or measurement variables have brought about the patterns in the observations.

These enable you to locate points or patterns of interest and identify their properties. You can also select individual components and highlight them for display to provide vitally different perspectives of your data.

The PCA Diagnostic Plots help you to assess the validity of the model via distance to model and modelling power plots.

4. Additional Capabilities

As well as the providing an analysis of Principal Components, CELLULA-P also comes packaged with a Nearest Neighbours tool to help identify the nearest. or similarly behaved, neighbours to a target observation.

You can read here about the full list of CELLULA-P features.