private static void AddFundAndFirmType(InvestmentActivity accession, InvestorHashTableSet investors)
{
if (string.IsNullOrEmpty(accession.ActivistInvestorName))
{
return;
}
var transformed = accession.ActivistInvestorName.Replace(" ", "").ToLower();
Investor possibleMatch = null;
if (investors.FirmNameAsKey.TryGetValue(transformed, out var firmNameMatch))
{
possibleMatch = firmNameMatch;
}
else if (investors.FundNameAsKey.TryGetValue(transformed, out var fundNameMatch))
{
possibleMatch = fundNameMatch;
}
if (possibleMatch != null)
{
accession.ActivistInvestorId = possibleMatch.Id.Substring(2); //CLEA 2.xxx system
accession.ActivistInvestorFundType = possibleMatch.FundType;
accession.ActivistInvestorFirmType = possibleMatch.FirmType;
return;
}
}
private static FileAnalysisResult AnalyzeFile(string file, IReadOnlyDictionary <string, InvestmentActivity> alteryxInvestmentSet,
IReadOnlyDictionary <string, int> supervised, InvestorHashTableSet investors)
{
InvestmentActivity investmentActivity = EdgarAnalyzer.Analyze(file);
Supervised trainingDatum = null;
Helper.MergeAccession(investmentActivity, alteryxInvestmentSet);
AddFundAndFirmType(investmentActivity, investors);
if (supervised.TryGetValue(investmentActivity.AccessionNumber, out var typeId))
{
trainingDatum = new Supervised(investmentActivity.AccessionNumber, investmentActivity.PurposeOfTransaction, typeId, investmentActivity.TypeOfReportingPerson);
investmentActivity.PurposeAnalyzedByML = false;
investmentActivity.PurposeOfTransactionTypeId = typeId;
}
return(new FileAnalysisResult()
{
Investment = investmentActivity, Supervised = trainingDatum
});
}
private static FileAnalysisBatchResult AnalyzeSliceOfFiles(int start, int limit, IReadOnlyList <string> files, IReadOnlyDictionary <string, InvestmentActivity> originAccessions,
IReadOnlyDictionary <string, int> supervised, InvestorHashTableSet investors, AtratinusConfiguration config)
{
IList <InvestmentActivity> investments = new List <InvestmentActivity>();
IList <Supervised> trainingData = new List <Supervised>();
var report = new FullQualityReport();
for (int fileIndex = start; fileIndex < limit; fileIndex++)
{
var analysis = AnalyzeFile(files[fileIndex], originAccessions, supervised, investors);
var fileReport = QualityGate.Measure(analysis.Investment, config);
report.ConsiderQualityReport(files[fileIndex], fileReport);
if (!fileReport.ShouldBeConsidered)
{
continue;
}
if (analysis.Supervised != null)
{
trainingData.Add(analysis.Supervised);
}
investments.Add(analysis.Investment);
}
return(new FileAnalysisBatchResult()
{
Accessions = investments, TrainingData = trainingData, QualityReport = report
});
}
