sLogicAnalyser comprises of state-of-the-art ready-to-use advanced algorithms for component analysis, data search and classification of massive datasets.  It consists of many modules based on supervised learning and unsupervised learning methods, statistical and non-statistical algorithms, and graph theoretical methods.  In addition to the well-defined datasets, our product can handle incomplete and partially connected datasets for analysis.

Customers can extract hidden patterns, association or similarities within massive datasets by using the modules developed from the most commonly used unsupervised learning algorithms such as K-means clustering, Expectation-Maximization (EM) clustering and Principal Component Analysis (PCA).  In the case of datasets with large dimensions, these unsupervised algorithms can be considered as pre-processing stage for performing supervised algorithms in our product.  Our consultants will work with customers on selecting suitable algorithm(s) based on the datasets.

Many supervised learning algorithms are also available for developing models using training set.  Customers can use one of the most commonly used Naïve Bayes and Bernoulli classification algorithms using Gaussian distribution for independent features in datasets. Linear and logistics regression algorithms along with regularization techniques can be used classify features in datasets without overfitting the model.

The above mentioned algorithms are good candidates for analyzing non-sequential datasets.  Customers can take advantage of using Hidden Markov and Continuous Random Field algorithms for modeling time-series sequential datasets.  During data analysis, it is sometimes necessary to discriminate and penalize certain outlying data points for developing better models.  To support such practices during data analysis, Radius Basis Kernal Functions and Support Vector Algorithms have been built-in to the tool.

One of the non-linear statistical algorithm is neural networks inspired by key functions of the brain.  Customers can use MultiLayer Perceptor (MLP) module for performing non-linear analysis and classification of features in massive datasets.