# SOSstat Features

*SOSstat* is rich and versatile software. It brings together a large number of
statistical and engineering techniques in a simple and powerful environment. The
graphical interface is organized in three distinct parts:

- A spreadsheet to store, modify and reorganize data. The spreadsheet can import an excel workbook with all its grids.
- A window dedicated to the display of numerical and graphical results. Graphs are resizable, reorganizable and exportable.
- A report that can be exported to a word processor, where the user can compile the results, either step by step or by automating the processing.

One of the strengths of *SOSstat* is to provide comments for each analysis, in
order to facilitate the interpretation of the results.

## SOSstat - Statistics

**SOSstat** ® provides a wide range of statistical procedures for exploration and
inference.

### Descriptive statistics

- Histogram
- Boxplot
- Correlation matrix
- Population parameters
- Statistical summary function
- Density estimation
- To find out more ...

### Normality tests

- Kolmogorov test
- Anderson Darling test
- Shapiro-Wilk test
- Probability plot
- To find out more ...

### Regression

Simple regression with different models:

- Linear with variable transformation
- Quadratic
- Confidence interval and prediction interval
- Applied test on a theoretical model
- Robust regression
- Orthogonal regression
- To find out more ...

Multiple regression

- Calculation of residuals, Studentized and standardized residuals
- Cook's distances and leverage calculation
- Standardization of predictors for analyzing designs experiments
- Prediction tool for the search for an optimum
- To find out more ...

### Parametric tests

Tests applicable when populations is normally distributed

- All the tests that follow can be selected using a wizard that directs the user to the test most suited to his problem.
- Compare the average or the variance to a theoretical value
- Comparison of two samples: Student test, Welch test, Fisher test, average test for paired samples
- Comparison of n samples : Bartlett's test, analysis of variance (ANOVA)
- To find out more ...

### Nonparametric Tests

Tests to be used when populations do not follow normal distributions

- Centering a sample with a reference value: Sign test, wilcoxon rank signed test
- Centering two samples: Wilcoxon Mann Whitney Test
- Centering test for two paired samples: Sign test
- Dispersion test for two samples: Ansari-Bradley test
- Centering test for n samples : Kruskal-Wallis Test
- Dispersion test for n samples : Levene test
- Comparison of two distributions : Kolmogorov 2 samples test
- To find out more ...

### Proportion tests

- Comparison of proportion to a theoretical value (exact and asymptotic test)
- Comparison of a frequency to a theoretical value (Poisson distribution)
- Comparison of two proportions (Fisher's exact test), and asymptotic test (Normal distribution)
- Comparison of frequencies (Poisson distribution)
- Comparison of several proportions

### Equivalence tests

These tests are particularly interesting to demonstrate that experimental results are "equivalent", i.e. the difference between experimental results cannot exceed a specified value. These tests are notably used to carry out bio-equivalence

- Equivalence of averages (several variants of this test)
- Population equivalence (several variants of this test)
- To find out more ...

### Detecting outliers

Detection of atypical values likely to pollute experimental results.

- Grubbs, Tukey, MADe test - detecting outliers in a sample
- Cochran test - outlier detection in several samples
- Graphical test in Boxplot

### Statistical intervals

- Tolerance interval
- Prediction interval of a mean
- Prediction interval of individual values

### Multivariate analysis

- Principal component analysis
- Clustering using the Kmeans method and Gaussian mixtures

## SOSstat - Quality

**SOSstat** provides a toolbox to improve product quality and control processes.

### Statistical process control SPC

- Cartes de contrôle aux mesures (
`\overline{X}/R`

,`\overline{X}/S`

, Moyenne Mobile, Valeurs Individuelles, Petite Série) avec journal de bord - EWMA Control chart
- Control charts for attributes: p, np, c and u
- Measurement sampling plan (S method)
- Process capability ans performance study (AFNOR, QS9000, 6 Sigma)
- Capability study for non-normal process: normal folded distribution , location error distribution, log-normal, stratified populations
- Inertial capability study
- Assessment of capabilities for a range of characteritics
- Wizard for calculating limites of control charts for measurement
- Wizard for calculating sample sizes
- Plot operating characteristic curves of an average control chart
- Wizard for calculation of capability threshold
- Validation of tools and machines
- To find out more ...

### Reliability

- Modeling laws failure distribution with Weibull or exponential distribution
- Management truncated, censored and suspended trials
- Success test to validate a reliability level

### Metrology

- Measurement capability analysis (Charbonneau, GRR method X-bar /R and R)
- Measurement capability study with Crossed ANOVA et nested ANOVA
- Type 1 Measurement capability (Cg - Cgk)
- Tests of bias and linearity for measurement systems
- Attributive capability studies : probabilistic method, signal method or analytical method
- Linearity study of measurement

### Designs of experiment (DOE)

Design of experiments and analysis of results

- Orthogonal designs : Full factorial design, fractional factorial design
- Design of two-level fractional designs according to the Taguchi method
- Surface Design : Centered Composite, Box Behnken
- Plackett and Burman designs
- D-optimal design
- Desirability function to optimize multiple responses
- Calculation of signal / noise ratios for product plans (Robustness)
- To find out more ...

### Quality Assistants

- Design and Evaluation of Sampling Plans for Attributes and Measures
- Design of sampling plans with
*Limiting Quality* - Calculation of control charts limits
- Evaluating the effectiveness of average control charts
- Calculating the optimal sample size for a control chart
- Calculation of the Z indicator of the 6-Sigma projects
- Calculation of capacity thresholds for risk control
- Bidirectional relationship between capability indicators and non-conforming rates
- Design of a reliability run test

Standard Used : AIAG MSA, AIAG SPC, ISO 11462-1, NFX06031-1:4, NFX06034, ISO_2021247, ISO 21747, ISO_7870, ISOTR22514-4, ISOTR7871

**SOSstat** - Graphics

**SOSstat** ® Offers a wide range of graphical representations :

- Boxplot
- Histogram
- Run chart
- Cumulative distribution function
- Multivari plot
- Synthesis (or summary) graphs - Pareto, Bar Graph on sample parameters
- 2D graphs - Time Series - Scatter (linear scale / log scale)
- 3D Graphs - Scatter, Contours, Surface
- Bubble plot
- Graphs for contingency tables - Radar graphs, Bar Graph

**SOSstat** - Various Operations

### Import/export

- Import text files, CSV
- Import / Export excel files
- Import data from a database - SQL Queries
- Instrument Connectivity USB port or RS232 port, Automation of analysis
- Record data in XML format, to ensure data survivability
- Encrypted file protection for email exchanges
- Export reports in RTF, WORD, HTML format

### Filtering / Sorting

- Filtering data according to a criterion
- Retrieving Data from the grids