public static List <OddsCSVMatch> LoadMatchOddsList()
{
var rows = Filey.LoadLines("HistoricalOdds.csv");
var oddsList = rows.Skip(1).Select(CreateFromCsv).ToList();
return(oddsList);
}
public static bool SuccessConditionLadderPrint(List <double> predicted, List <double> actual, string print)
{
var phLadder = Numbery.Denormalise(predicted[0], Util.MaxLadderPoints);
var paLadder = Numbery.Denormalise(predicted[1], Util.MaxLadderPoints);
var ahLadder = Numbery.Denormalise(actual[0], Util.MaxLadderPoints);
var aaLadder = Numbery.Denormalise(actual[1], Util.MaxLadderPoints);
if (print == "")
{
Console.WriteLine("[{0}, {1} Vs {2}, {3}] Suggested Bet: ${8:0.00}", phLadder, paLadder, ahLadder, aaLadder);
}
else if (!string.IsNullOrEmpty(print))
{
Filey.Save(string.Format("[{0}, {1} Vs {2}, {3}] Suggested Bet: ${8:0.00}", phLadder, paLadder, ahLadder, aaLadder), print);
}
if (phLadder > paLadder && ahLadder > aaLadder)
{
return(true);
}
if (phLadder < paLadder && ahLadder < aaLadder)
{
return(true);
}
if (phLadder == paLadder && ahLadder == aaLadder)
{
return(true);
}
return(false);
}
public static bool SuccessConditionTotalPrint(List <double> predicted, List <double> actual, string print = "")
{
var phTotal = Numbery.Denormalise(predicted[0], Util.MaxScore);
var paTotal = Numbery.Denormalise(predicted[1], Util.MaxScore);
var ahTotal = Numbery.Denormalise(actual[0], Util.MaxScore);
var aaTotal = Numbery.Denormalise(actual[1], Util.MaxScore);
if (print == "")
{
Console.WriteLine("[{0}, {1} Vs {2}, {3}]", phTotal, paTotal, ahTotal, aaTotal);
}
else if (!string.IsNullOrEmpty(print))
{
Filey.Append(string.Format("[{0}, {1} Vs {2}, {3}]", phTotal, paTotal, ahTotal, aaTotal), print);
}
if (phTotal > paTotal && ahTotal > aaTotal)
{
return(true);
}
if (phTotal < paTotal && ahTotal < aaTotal)
{
return(true);
}
if (phTotal == paTotal && ahTotal == aaTotal)
{
return(true);
}
return(false);
}
public TwitterHelper()
{
//TODO: should pull this from a Keystore
var file = Filey.Load(@"C:\Users\Jack\Documents\Visual Studio 2012\Projects\AFLTipper\keys.keys").Split('\n');
APIKey = file [0].Replace("API key: ", "").Replace("\r", "");
APISecretKey = file[1].Replace("API secret key: ", "").Replace("\r", "");
BearerToken = file[2].Replace("Bearer token: ", "").Replace("\r", "");
AccessToken = file[3].Replace("Access token: ", "").Replace("\r", "");
AccessTokenSecret = file[4].Replace("Access token secret: ", "").Replace("\r", "");
}
public void FindMe_Should_FindDetailsInSafewayHTML()
{
//Arrange
string path = AppDomain.CurrentDomain.BaseDirectory;
var html = Filey.Load(path + @"\..\..\TestDataSafewayHTML.txt");
//HtmlySearchInformation searchInformation = new HtmlySearchInformation();
//Act
var cost = Htmly.FindCostIn(html, "almonds");
//Assert
Assert.IsTrue(cost.Length > 0);
}
public static bool SuccessConditionLadderGoalsAndPointsPrint(List <double> predicted, List <double> actual, string print)
{
Func <double, double> rule = (m => m > 27.0 ? 15.00 : 0.00);
var phLadder = Numbery.Denormalise(predicted[0], Util.MaxLadderPoints);
var paLadder = Numbery.Denormalise(predicted[1], Util.MaxLadderPoints);
var phGoals = Numbery.Denormalise(predicted[2], Util.MaxGoals);
var phPoints = Numbery.Denormalise(predicted[3], Util.MaxPoints);
var paGoals = Numbery.Denormalise(predicted[4], Util.MaxGoals);
var paPoints = Numbery.Denormalise(predicted[5], Util.MaxPoints);
var phScore = phGoals * 6 + phPoints;
var paScore = paGoals * 6 + paPoints;
var ahLadder = Numbery.Denormalise(actual[0], Util.MaxLadderPoints);
var aaLadder = Numbery.Denormalise(actual[1], Util.MaxLadderPoints);
var ahGoals = Numbery.Denormalise(actual[2], Util.MaxGoals);
var ahPoints = Numbery.Denormalise(actual[3], Util.MaxPoints);
var aaGoals = Numbery.Denormalise(actual[4], Util.MaxGoals);
var aaPoints = Numbery.Denormalise(actual[5], Util.MaxPoints);
var ahScore = ahGoals * 6 + ahPoints;
var aaScore = aaGoals * 6 + aaPoints;
var margin = Math.Abs(phScore - ahScore);
var wager = rule(margin);
if (print == "")
{
Console.WriteLine("[{0}: {1}, {2}: {3} Vs {4}: {5}, {6}: {7}] Suggested Bet: ${8:0.00}", phLadder, phScore, paLadder, paScore, ahLadder, ahScore, aaLadder, aaScore, wager);
}
else if (!string.IsNullOrEmpty(print))
{
Filey.Save(string.Format("[{0}: {1}, {2}: {3} Vs {4}: {5}, {6}: {7}] Suggested Bet: ${8:0.00}", phLadder, phScore, paLadder, paScore, ahLadder, ahScore, aaLadder, aaScore, wager), print);
}
if (phScore > paScore && ahScore > aaScore)
{
return(true);
}
if (phScore < paScore && ahScore < aaScore)
{
return(true);
}
if (phScore == paScore && ahScore == aaScore)
{
return(true);
}
return(false);
}
public static bool SuccessConditionTotal(List <double> predicted, List <double> actual)
{
var phTotal = Numbery.Denormalise(predicted[0], Util.MaxScore);
var paTotal = Numbery.Denormalise(predicted[1], Util.MaxScore);
var ahTotal = Numbery.Denormalise(actual[0], Util.MaxScore);
var aaTotal = Numbery.Denormalise(actual[1], Util.MaxScore);
Filey.Append(string.Format("[{0}, {1} Vs {2}, {3}]", phTotal, paTotal, ahTotal, aaTotal), @"C:\Users\Jack\Desktop\ANN Optimizer results.txt");
if (phTotal > paTotal && ahTotal > aaTotal)
{
return(true);
}
if (phTotal < paTotal && ahTotal < aaTotal)
{
return(true);
}
if (phTotal == paTotal && ahTotal == aaTotal)
{
return(true);
}
return(false);
}
public override void Train(Network network, List <List <double> > inputs, List <List <double> > targets)
{
network.Epochs = 0;
var minima = 0;
double minError = -1;
double maxError = -1;
double prevError = -1;
const double cycleLength = 100;
double cullPopulation = 50;
var log = new List <List <double> >();
do
{
network.Error = network.TrainEpoch(inputs, targets) / inputs.Count;
network.Epochs++;
minima++;
if (network.Error < minError || minError < 0)
{
minima = 0;
minError = network.Error;
//Network.Save(network); //TODO: reinstate using mongodb
}
if (network.Error > maxError)
{
maxError = network.Error;
}
if (network.Error > prevError)
{
AdjustLearningRateDown(network);
}
prevError = network.Error;
log.Add(new List <double>()
{
network.Epochs, minima, network.Error, minError, maxError
});
//Cull outliers
var largestError = 0;
var populationErrors = new Dictionary <int, double>();
var temp = "Number,Error\n";
if (network.Epochs % cycleLength == 0)
{
//for each of the training cases
for (var i = 0; i < inputs.Count; i++)
{
//get the output List for the given input List
var outputs = network.Run(inputs[i]);
var error = outputs.Select((t, j) => Math.Abs(targets[i][j] - t)).Sum();
populationErrors.Add(i, error);
}
temp = populationErrors.Aggregate(temp, (current, p) => current + ("" + p.Key + "," + p.Value + "\n"));
Filey.Save(temp, "tamp_ANNOutput.txt");
var boundary = LocatePercentileError(0.995, populationErrors.Select(e => e.Value).ToList());
var cullInputs = populationErrors.Where(p => p.Value > boundary).Select(e => inputs[e.Key]);
var cullTargets = populationErrors.Where(p => p.Value > boundary).Select(e => targets[e.Key]);
foreach (var i in cullInputs)
{
inputs.Remove(i);
}
foreach (var t in cullTargets)
{
targets.Remove(t);
}
}
} while (minima < network.MaxMinima && network.Epochs < network.MaxEpochs);
network = Network.Load(network.Directory + network.Id + ".ann");
Filey.Save(log, "Network/Algorithm/Log.txt");
var rankings = network.RankInputs();
Filey.Save(rankings, "Network/Algorithm/Rankings.txt");
}