Benefactors:

Aims:

Essential: Develop a web tool (interface to a model), capable of allowing people to:
- Allow the user to look back at past events, see what the actual footfall was, and what our model predicted it would be under normal conditions.
- Allow the user to ask for estimates of future footfall (they will need to enter all of the required inputs (e.g. temp, rain, etc.)).
Would be nice: [xx] - See if a Bayesian method preforms better than random forest [MA]: The Bayesian method does not provide an improved prediction over Random Forest. The accuracy of Bayesian method is very similar to that of Linear Regression. Therefore, )
- The model automatically scrapes new data and retrains itself to make better future predictions

Collect up-to-date version of the input data and make a new input data file
Re-implement the Random Forest model in R
Check that the new model makes similar estimates to the Python one
Create an R Shiny app capable of:
Allowing the the user to run the model on historical data to see what the real footfall v.s. prediction was (e.g. a Retrospective Analysis tab).
Allowing the user to use the model to make predictions of the future footfall, given their estimates of the weather conditions etc. (e.g. a Prospective Analysis tab).

Others