This is a simple interactive tool for preference analysis based on stated preference method (SP) and discrete choice models (DCM).
A series of virtual experiments regarding home location choice would be given to participants. In each choice situation, a participant
will compare between two alternatives, the left one is supposed to be his/her current residence, and the right one is a new residence.
Both of them are described by 5 attributes, namely commute distance, size, monthly rent, population density of the community,
and income structure of the community. participants are encouraged to do as many experiments as possible, but they can stop at any time
by clicking "Finish" button on the bottom to run the preference analysis.

The preference analysis is implemented by DCM with 3 options. The first option is to run a simple logit model on the participant's
own data, and thus the results would be called ego model. However, please note that it would be unstable in many cases when the
participant only played with a few number of experiments, one will find the coefficients are too large, and the goodness-of-fit (r2) is
unrealistically closed to 1. The second option is also to run a simple logit model, but on all data. The results indicate the mean
preference of the whole population. This so-called population model might be the most practical one with robustness and simplicity. The
third option is to run a mixed logit model (or called random parameter model), which is much more complicated than the simple logit. The
greatest advantage of this model is that each coefficient is a distribution, rather than a fixed number. In this way, the participant
can check the diversity of preference among population, and learn what position his/her individual preference lays in the distribution.
To better understand the results, some explanative notes are avaiable for each option.

After the preference analysis, the participant can go back to playground of virtual experiments and check the predictions of the model
on each choice situation by clicking "Predict" at the bottom.