Our company uses several mining and modelling techniques. Complex problems require the usage of multiple techniques. MicroSegment developed the MMSegment methodology that incorporates and provide guide the application of standard modelling techniques like:
- Most commonly used algorithms for predictive modeling methods are decision trees, neural networks, and statistical models.
- Clustering models focus on identifying groups of similar records. Records in the same cluster have similar attributes, while the distance between groups are high and the similarity is low (Neural Networks, Kohonen Networks, K-Means)
- Association rules are useful to create easily interpretable rules to reach particular conclusions (such as the purchase of a particular product).
- Text mining techniques would allow to analyze unstructured information