static IList <Entity> FilterTransactions(IEnumerable <Transaction> transactions, string description, double threshold)
{
// Load up our dictinoary of canonized names
var wordParser = new KeywordParser("Data/Aliases.csv");
var keywords = wordParser.Parse(description);
var matches = new List <Transaction>();
Console.WriteLine("Filtering transactions...");
foreach (var item in transactions)
{
if (String.IsNullOrWhiteSpace(item.Name))
{
continue;
}
var words = wordParser.Parse(item.Name);
if (words.Length > 0 && words.KeywordMatch(keywords) >= threshold)
{
matches.Add(item);
}
}
var results = new List <Entity>();
Console.WriteLine("Aggregating transactions...");
var groups = matches.GroupBy(i => i.filer_id);
foreach (var group in groups)
{
var entity = new Entity
{
Keywords = keywords,
Transactions = group.ToList()
};
results.Add(entity);
}
#if DEBUG
DumpEntities(results);
#endif
return(results);
}
static List <Entity> AggregateTransactionsByDonor(IEnumerable <Transaction> transactions)
{
var results = new List <Entity>();
var count = transactions.Count();
int i = 0;
// Load up our dictinoary of canonized names
var wordParser = new KeywordParser("Data/Aliases.csv");
using (var progress = new ProgressBar())
{
Console.Write("Aggregating transactions...");
foreach (var item in transactions)
{
var words = wordParser.Parse(item.Description);
if (words.Length > 0)
{
var entity = BestEntityMatch(results, words, 0.8);
if (entity == null)
{
// If no good entity match, create a new one
entity = new Entity
{
Keywords = words,
Transactions = new List <Transaction>()
};
results.Add(entity);
}
entity.Transactions.Add(item);
}
progress.Report((double)++i / count);
}
Console.WriteLine("Done.\n");
#if DEBUG
DumpEntities(results);
#endif
}
return(results);
}