Comment from Marcus

[fa-mc]

This is a wonderful idea, also very ambitious. It tries to use the population profile, city development plan, location, interest rate and policy change to predict property prices in the coming years. However, it would be extremely complicated to implement due to the large amount of assumptions and predictions needed as the inputs to make the final prediction.

My suggestion is to build a base case model first, using variables such as population profile, city development plan and location. Those variables are relatively stable (You wouldn’t expect a large shift in population profile or city development plan in a short period of time). Other factors, such as interest rate and policy, are be extremely difficult to predict. It would take significant amount of resources (e.g. academic training, research and data collection) just to get a very rough (also subjective) estimation of the future state. Using these predictions as inputs of your model would make the results very volatile. Given the cost and difficulty, it would be better to just assume Status Quo.

However, you could add a layer of scenario test over the basic model. For example, you can include fields of “what if” conditions and let the users to play with it (E.g. what if the interest rate increases by 2%, or what if a foreign buyer tax is enforced). In that way, you can avoid the trouble of predicting these status, but still provide users the ability to tweak the model to reflect their own view of the future.

Overal it’s a very interesting project. I recommend you reduce the complexity to make it easier for implementation.

[aassdd654]

Hi Marcus, Thanks a lot for your advice, your comment (base model/what issue scenario structure)gave me a practical start of the project! Appreciated!