Calulate probability of points allocation

[Sephiroth100]

Hi,

First time posting so be gentle :)

I have a website and on the website I calculate popularity of objects over time.

One of the objects is called 'Store Category' - a store has one or more categories.

I.e. the store Easyjet, has categories 'Flights', 'Hotels' etc.

In order to apply popularity at this level the visitor must have first chosen a category. This is not always the case, because I might present a list of stores and the visitor selects one directly, without narrowing the stores by category first.

When this happens I'm left with a pot of popularity points where the category could not be identified. I would like to try and allocate the points using some kind of algorithm.

So, take our example of Easyjet

Easyjet operates in 3 categories, 'Flights', 'Hotels' and 'Travel Insurance'.

Each category is operational for x days and has accrued x identifiable points.

Flights - 90 days, 180pts
Hotels - 30 days, 60pts
Travel Insurance - 60 days, 0pts​

I have 180pts in the pot that are yet to be allocated across the 3 categories, but what's a fair assumption?

My first thoughts were a combination using time and points independently because otherwise categories with no points would never get allocated points from the pot.

points to allocate = pot points * (days/total days + pts/total pts / 2)
using Flights: 105 = 180 * ((90/180 + 180/270) / 2)​

But I'm no mathematician so any help would be appreciated.

Thanks

John

 

[AndyW]

Dear John,

Thank you for sharing your question with us. It sounds like you are facing a common challenge in data analysis, where some data points are missing or incomplete. In order to address this issue, there are a few things you could consider.

One option is to use a technique called imputation, which involves filling in missing values with estimated values based on the available data. In your case, you could use the available data on the popularity of the categories to estimate the missing values for the category that was not chosen by the visitor. This could be done by calculating the average popularity score for each category and using that as the estimated value for the missing data. This approach would help to distribute the popularity points fairly among the categories.

Another option is to use a regression analysis to predict the popularity of the category based on the available data. This would involve creating a model that takes into account factors such as the number of days the category has been operational and the number of identifiable points it has accrued. The model could then be used to estimate the missing popularity points for the category that was not chosen by the visitor.

Ultimately, the best approach will depend on the specifics of your data and your research question. I would recommend consulting with a statistician or data analyst to determine the most appropriate method for your particular situation.

I hope this helps and good luck with your analysis!