SIMALTO Modelling

Conventional regression methods are not appropriate for SIMALTO data (non linear variables, correlated variables etc.) and so Expert systems and Genetic algorithm approaches are used which parallel the thinking behind Neural Networking and Data Mining techniques.
All analysis is done at the individual respondent level – there is no ‘averaging’.

The unique simulation reports are :-

Optimum specification at ANY total price/cost constraint. This is the specification (one option selected on each attribute) that maximises respondent preference – i.e. provides as many respondents as possible with as many of their high priority choices and as few of their low priority choices as possible, for a given total cost/price.

Value Preference Higherarchy. The preference of any particular option over other options of that attribute, recognising that ‘better’ options tend to ‘cost more’.

Needs-based cluster analysis. Finds those respondents that have similar priorities and separates them from groups of respondents with a different set of priorities.

Preference (market) share analysis. The user inputs different specifications (and brand and price if appropriate) and the modelling predicts preference share, all other things being equal.

Activity forecasts. Assuming the appropriate ‘variable’ question is asked in the SIMALTO interview, the models can predict the likely respondent action on this ‘variable’ scale for any product/service specification e.g. would they use the product more, or be more ‘loyal’, or pay more for it and if so how much, etc. etc.