The Unscrambler®: Key Features
| Features | ||
 |
||
| What New? | ||
|
|
|
- Descriptive statistics (Mean, Standard Deviation, Box-Plot, skewness, kurtosis, correlation matrix)
- Principal Component Analysis (PCA)
- Multivariate Curve Resolution (MCR)
- Regression (MLR, PCR, PLS-R, 3-way PLS-R)
- Prediction from PCR, PLS-R and 3-way PLSR models
- Clustering (K-Means)
- Classification (SIMCA, PLS-DA)
- ANOVA and Response Surface ANOVA
- Validation options: Leverage Correction, Cross-Validation (freely choose number of samples per segment), Test Set
- Variable Scaling options: Scaling is free on each variable. Suggested options: Auto-scaling, Constant, Passify
- Interaction and Square terms can be included in PCA, MLR, PCR and PLS-R models
| Top |
- Automatic detection of significant X-variables in PCR and PLS-R
- Model Stability in PCA, PCR and PLS-R
- Automatic outlier detection in PCA, MLR, PCR, PLS-R and Prediction
- Message list of recommendations in MCR modeling
- Interactive analysis:
- Mark samples and/or variables on plots
- Recalculate With or Without Marked samples or variables
- Recalculate With Passified Marked or Unmarked variables
- Extract Data from Marked or Unmarked
- Automatic Pretreatments in Prediction and Classification
- Interactive help
- Tutorial exercises guiding you through the use of all The Unscrambler® modeling techniques in application examples
| Top |
- Design Wizard: Takes you through the stages of building a design
- Fractional and Full Factorial designs
- Plackett-Burman designs
- Central Composite designs
- Box Behnken designs
- Mixture designs (Simplex-Lattice, Axial, Simplex-Centroid)
- D-Optimal designs of mixtures and non-mixtures
- Effects plots of main effects and interaction effects; response surface plots
| Top |
- Smoothing: Moving Average, Savitzky Golay, Median filter, Gaussian filter
- Normalize: Area, Unit Vector, Mean, Maximum, Range, Peak
- Spectroscopic conversions: absorbance/reflectance, reflectance/ Kubelka-Munk, wave number / nanometers
- Multiplicative Signal Correction (MSC) & Extended MSC (EMSC)
- Noise
- Derivatives: Norris-Gap, Gap-Segment and Savitzky Golay
- Baseline offset and Linear Baseline correction
- Standard Normal Variate (SNV)
- Mean centering, standard deviation scaling
- User-Defined pretreatments (programmed for e.g. in Matlab or C++ and utilized in The Unscrambler® as a dll)
| Top |
- Over 100 pre-defined plots (Scores, Loadings, Loading weights, Correlation Loadings, Stability Plot, Bi-plot, Leverages, Variances, RMSE, Regression coefficients, Predicted vs. Measured, Residuals, Response Surface…)
- Many plot options: Sample grouping with colors or symbols; Insert text; Remove grid; Display plot as Line/Bars/Accumulated bars…
- Live rotation of 3D plots
- PC Navigation tool: navigate along model components in a single click
| Top |
- No size limit
- 2D and 3D tables supported
- Continuous, discrete and mixture variables supported
- Sample and variable names of up to 49 characters
- Raw data plots (Line, 2D scatter, 3D scatter, Normal probability, Histogram, Matrix, Matrix 3D)
- Set Editor: allows to work freely either on the whole table or on a selection of samples and/or variables
- Sort samples / Sort Variables
- Transpose
- Mathematical array manipulation
- Fill missing values
- Undo/Redo functionality
| Top |
Interactions with Other Software
- Drag 'n' drop values, sample names and variable names from Excel to The Unscrambler®
- Copy/Paste data or plots to Microsoft Office applications
- Data import of 2D and 3D data tables with over 20 file formats supported: ASCII, spreadsheet software (Excel, Lotus), statistical software (NSAS, Matlab) and instrument software: JCAMP-DX, Tracker (Foss), APC, Grams (Thermo Electron), MVACDF, Indico (Analytical Spectral Devices), CLASS-PA & SpectrOn (Guided-Wave), F3D (Hitachi), BFF3 (Brimrose)
- User-Defined Import: allows to import any file format using a preprogrammed DLL
- Data export in various formats (ASCII, Matlab, JCAMP-DX…)
- Model export in various formats (ASCII, NSAS, Vision, Tracker)
- Password protected Login, with password expiration after 70 days
- Secure electronic signatures
- The user name, date, time and software version used to generate data and model files are registered
- Details from the instrument source file are passed on into the Unscrambler data file under importation
- All formatting operations on the data set are recorded in the Log trail
- All computations performed on the data set are recorded in the Log trail
- All transformations performed on the data set are recorded in the Log trail
- Locked data sets cannot be edited
- The source data set’s name and path are recorded in the resulting model file
- The source data set’s transformations are recorded in the resulting model file
- The complete and detailed model settings are recorded in the resulting model file
- Printed plots contain date, time, file name, software name and user initials
Sample grouping makes it easier to distinguish between groups or classes of samples plotted directly from the Unscrambler® editor, especially useful for spectroscopic users.
Easy activation of pre-treatment steps
Auto-pre-treatment registration functionality made simple to use.
The philosophy of The Unscrambler® is focused on projection methods, and especially PLS regression. MLR prediction is now available in OLUP. MLR is in use by (but not limited to) spectroscopic users.
File Merge (PCA, PLS1, PLS2, PCR, MLR, MSC, EMSC)
The Unscrambler® file system has a set of files to represent the model and data file. When a model file is saved in The Unscrambler® it produces 5 files with extensions of *.xxD, *.xxP, *.xxL, *.xxT, *.xxW. The main information of the file is in *.xxD. Other information about the model like notes, log information and history of the file is distributed across the sub files.
File merge option now merges all the files into one single file. This single file consists of all the sub files and also the MSC and EMSC model files if the pre-treatments are used in the model file.
3rd Order Variable Interactions
Third order variable interactions include interactions between 3 design parameters (A*B*C), cubic terms of the design variables (A3) and combined effects (A2*B).
R2 (read: R-square) is now available in addition to R in the statistics box in the predicted vs measured plot.
The Unscrambler® 9.7 offers four different rotations: Varimax, Quartimax, Equimax and Parsimax. These rotations are to be applied after PCA analysis.
