# Load necessary libraries
library(data.table)
library(dplyr)

# This function can be used to calculate your prediction scores before submitting to Kaggle
evaluate = function(finalPredictions){
    
    # Load the data
    tourneyCompact = fread("../input/TourneyCompactResults.csv")
    
    # Filter data for seasons: 2013, 2014, 2015 & 2016
    seasons2Test = seq(2013, 2016)
    tourneyCompact = filter(tourneyCompact, Season %in% seasons2Test)
    
    # Remove the Play-In ("First Four") games
    tourneyCompact = filter(tourneyCompact, Daynum > 135)
    
    # Select only relavent columns
    tourneyCompact = tourneyCompact[, c(1, 3, 5)]
    tourneyCompact$Result = 1
    names(tourneyCompact)[c(2:3)] = c("Team1", "Team2")
    
    # Swap the teams and change the result if the winning team number is 
    # greater than the losing team number
    tourneyCompact$Result = as.numeric(!(tourneyCompact$Team1 > tourneyCompact$Team2))
    tourneyCompact$temp = 0
    tourneyCompact$temp[tourneyCompact$Result == 0] = tourneyCompact$Team1[tourneyCompact$Result == 0]
    tourneyCompact$Team1[tourneyCompact$Result == 0] = tourneyCompact$Team2[tourneyCompact$Result == 0]
    tourneyCompact$Team2[tourneyCompact$Result == 0] = tourneyCompact$temp[tourneyCompact$Result == 0]
    tourneyCompact = tourneyCompact[, -5]
    
    # Create a dataset with "id" and "Result" variables
    actualResults = tourneyCompact %>% mutate(id = paste(Season, Team1, Team2, sep = "_"))
    actualResults = select(actualResults, id, Result)
    
    # Merge our predictions with actual results to form a "final" dataset
    finalPredictions$id = as.character(finalPredictions$id)
    finalPredictions$pred = as.numeric(as.character(finalPredictions$pred))
    final = merge(actualResults, finalPredictions)
    
    # Calculate the log loss for each match
    final = final %>% mutate(predProb = (Result*log(pred) + (1-Result)*log(1-pred)))

    # Calculate our final score
    score = -sum(final$predProb)/nrow(final)
    return (score)
}

# SampleSubmission
#sampleSubmission = fread("../input/sample_submission.csv")
#evaluate(sampleSubmission)