The Unscrambler® X Features

Exploratory Data Analysis

Descriptive Statistics

Mean/ Std Dev/ Quartiles/ Cross Correlations/ Scatter Effects

Statistical Tests

Normality Test/ t-Tests/ F-Tests/ Mardia’s Multivariate Test

Cluster Analysis

K-means / K-medians Ward’s method

Hierarchical Cluster Analysis (HCA) with dendrograms

Principal Component Analysis (PCA)

Choice of using NIPALS or SVD algorithms

Rotation methods including Varimax, Equimax, Quartimax and Parsimax

Multivariate Curve Resolution (MCR)

Resolve time evolving data such as chemical reaction or chromatographic data into pure constituent profiles and pure spectra

Regression and Classification

Regression Methods

Multiple Linear Regression (MLR) / Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR)

Choice of algorithms, NIPALS and SVD for PCR and NIPALS, Kernel Methods and Orthogonal Scores for PLSR
Improved Test Set Validation options

Q-Residuals in influence plots
L-PLS, incorporating three data tables for greater insights into data structure

Advanced Classification Methods

Projection using PCA and PLS models

Soft Independent Modelling of Class Analogy (SIMCA) now also incorporating PLS models

Linear Discriminant Analysis (LDA)

Support Vector Machines (SVM) Classification with numerous kernel types

Data Pretreatments

Spectral Functions


Derivatives: Moving Average/ Norris Gap/ Savitzky-Golay

Baseline Correction


Spectroscopic: Reflectance/ Transmission/ Kubelka-Munck

Scatter Correction and Advanced Functions

Multiplicative and Extended Multiplicative Scatter Correction (MSC/ EMSC)

Standard Normal Variate (SNV)



General Transforms

Improved Centre and Scale options

Spectroscopic: Reflectance/ Transmission/ Kubelka-Munck

Interaction & Squares and Individual Variable Weighting

Compute General

Fill Missing Values

Correlation Optimization Warping (COW)


Design of Experiments

Hard-to-change factors handled via split plots

Two-level, general and optimal factorial split-plot designs

Half-normal selection of effects from spiltplot experiments with test matrices that are balanced and orthogonal

Power calculated for spilt plots versus the alternative of complete randomization

Improved capabilities to confirm or verify model predictions

Entry fields for confirmation data and calculation of mean results

Enter verification runs embedded within blocks as controls or appended to your completed design

Verification points displayed on model graphs and raw residual disgnistics

Flexibility in data display and export

Journal feature to export data directly to Microsoft Word or Powerpoint

Copy/paste of Final Equation from ANOVA to Microsoft Excel

New XML script commands for exporting point predictions Interactive Tables

New design and graphics capabilities

Definitive screening designs

Select a simple-sample design for mean-model only

One-sided option added to FDS graph Interactive Tables

Enter a single factor constraint for response surface designs

Adjustably-tuned LOESS fit line for Graph Columns