base line predictor

baseline.fs

 //Non personalized baseline predictor 
    let baseline (ratings:(float list)list) = 
        let mu = ratings |> List.map ( fun ra -> [for i in 0 .. ra.Length - 1 -> ra.[i]] 
                                                 |> List.filter (fun t -> t <> 0.0) 
                                                 |> List.average) 
                         |> List.average         
        let mutable bu = ratings |> List.sumBy (fun rating -> [for i in 0 .. rating.Length - 1 -> rating.[i]] 
                                                              |> List.filter (fun ri -> ri <> 0.0) 
                                                              |> List.sumBy (fun ri -> ri - mu))  
                 
        bu <- bu / float ratings.[0].Length 
        let mutable bi = ratings |> List.sumBy (fun ra -> [for i in 0 .. ra.Length - 1 -> ra.[i]] 
                                                          |> List.filter (fun t -> t <> 0.0) 
                                                          |> List.sumBy (fun z -> z - bu - mu))  

        bi <- bi / float ratings.Length 
        mu + bu + bi 
     

I got here from the link in your "F# Machine Learning Essentials" book. However, you don't explain this formula neither in the book nor in this code. It would be very nice if you could explain the calculations. Thanks.