public override ContinuousDistribution Plus(ContinuousDistribution two)
{
if (two is ClippedGaussianDistribution)
{
ClippedGaussianDistribution other = (ClippedGaussianDistribution)two;
return(new ClippedGaussianDistribution(mean + two.Mean, variance + two.Variance, lower + other.lower, upper + other.upper));
}
return(new GaussianDistribution(mean + two.Mean, variance + two.Variance));
}
public IDataSource<string, ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>> ImputeEmotionalContentFromFile(string filename, uint column, uint repeats, string imputesave)
{
MemorySource<string, ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>> imputed = new MemorySource<string, ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>>();
ComboSource<string, ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>> combo = new ComboSource<string, ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>>(source, imputed);
// Check for existing imputed file
DataReader imputereader = new DataReader(imputesave);
uint kk = 0;
for (string[] row = imputereader.ReadRow(); row != null; row = imputereader.ReadRow()) {
kk++;
if (kk % 1000 == 0)
Console.WriteLine("#" + kk);
double meanv = double.Parse(row[1]), varv = double.Parse(row[2]), meana = double.Parse(row[3]), vara = double.Parse(row[4]), meand = double.Parse(row[5]), vard = double.Parse(row[6]);
ContinuousDistribution valence = new ClippedGaussianDistribution(meanv, varv, 0, 1);
ContinuousDistribution arousal = new ClippedGaussianDistribution(meana, vara, 0, 1);
ContinuousDistribution dominance = new ClippedGaussianDistribution(meand, vard, 0, 1);
imputed[row[0]] = new ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>(valence, arousal, dominance);
}
imputereader.Close();
for (uint ii = 0; ii < repeats; ii++) {
Dictionary<string, List<KeyValuePair<double, double>>>
sentencesV = new Dictionary<string, List<KeyValuePair<double, double>>>(),
sentencesA = new Dictionary<string, List<KeyValuePair<double, double>>>(),
sentencesD = new Dictionary<string, List<KeyValuePair<double, double>>>();
DataReader reader = new DataReader(filename);
uint jj = 0;
for (string[] row = reader.ReadRow(); row != null; row = reader.ReadRow()) {
jj++;
if (jj % 1000 == 0)
Console.WriteLine("#" + jj + ": " + sentencesV.Count + ", " + imputed.Count);
List<string> words = TwitterUtilities.SplitWords(row[column].ToLower());
AnalyzeWords(words, combo, sentencesV, sentencesA, sentencesD);
}
reader.Close();
AnalyzeSentences(imputed, sentencesV, sentencesA, sentencesD, imputesave);
}
source = combo;
return imputed;
}
public void AnalyzeSentences(MemorySource<string, ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>> inputed,
Dictionary<string, List<KeyValuePair<double, double>>> sentencesV,
Dictionary<string, List<KeyValuePair<double, double>>> sentencesA,
Dictionary<string, List<KeyValuePair<double, double>>> sentencesD, string imputesave)
{
using (var stream = File.CreateText(imputesave)) {
foreach (string stem in sentencesV.Keys) {
double vmean = WeightedStatistics.Mean(sentencesV[stem]), amean = WeightedStatistics.Mean(sentencesA[stem]), dmean = WeightedStatistics.Mean(sentencesD[stem]);
ClippedGaussianDistribution valence = new ClippedGaussianDistribution(vmean, WeightedStatistics.Variance(sentencesV[stem], vmean, true), 0, 1);
ClippedGaussianDistribution arousal = new ClippedGaussianDistribution(amean, WeightedStatistics.Variance(sentencesA[stem], amean, true), 0, 1);
ClippedGaussianDistribution dominance = new ClippedGaussianDistribution(dmean, WeightedStatistics.Variance(sentencesD[stem], dmean, true), 0, 1);
inputed[stem] = new ThreeTuple<ContinuousDistribution, ContinuousDistribution, ContinuousDistribution>(valence, arousal, dominance);
stream.WriteLine(stem + "," + valence.InternalMean + "," + valence.InternalVariance + "," + arousal.InternalMean + "," + arousal.InternalVariance + "," + dominance.InternalMean + "," + dominance.InternalVariance);
}
}
}
