statmining isoft

The number one tool designed to industrialise your statistical processes

Data mining has taken off in a big way since the late 1990s thanks to the development of specialist software tools for "excavating" data. Interfacing statistical algorithms has helped to spread the implementation and use of statistical models (scores).

For a company to gain market share today, it has to offer its customers made-to-measure products. The significant growth in commercial opportunities brings rewards provided a business knows how to industrialise the creation and maintenance of its statistical models and analyses.


Carrying out an analysis or model is an iterative process that ranges from acquiring data to modifying and modelling it. For the procedure to be effective, users must be able perform the various steps in the same work environment.

data management isoft

A breakthrough innovation: data morphing

Photographers use image morphing to view the alterations made to an image in real time by applying filters, distortions or other processes.

ISoft’s real-time calculation engine set out to reproduce the same calculation interactivity in StatMining. For each data transformation (discretisation, replacing missing values, selecting records, etc.), the user sees the results of the operation instantly on billions of records. This breakthrough innovation makes StatMining a unique productivity tool.

data morphing isoft
méthode statmining isoft


StatMining integrates our tried and tested methodology for analysing and creating statistical models and developing machine-learning algorithms.
StatMining consists of the following stages:

  • Collect: importation, quality control, partitioning, simple sampling, clusters, etc.
  • Explore: frequency tables, contingency, χ2, distribution of variables, correlation, etc.
  • Prepare: manual, automatic and supervised discretisation, outlier detection, replacement of missing values, etc.
  • Analyse: factorial analyses - PCA, MCA, classification of individuals, variables, etc.
  • Model: machine-learning algorithms, LARS, decision trees, forward-backward-stepwise field selection methods, etc.
  • Validate: score performance tables, re-application of models, etc.
  • Exploit: production launch, production monitoring, model updating, etc.


StatMining is a multi-sector software solution dedicated to the industrial production of analyses and models. It can be used to solve the following types of question:

  • Which product for which customer?
  • What is the behaviour of my customers?
  • What is the risk associated with each of my clients?


assurance santé isoft
Banking, insurance (life, property, casualty and health insurance)
distribution statmining isoft
Retail, mail order
énergie isoft
Telephone market, public sector, energy