public PatternMiningResult FilterRules(
PatternMiningResult results,
int minimumRuleSupport = 500,
double minimimRuleOccFraction = 0.7)
{
var filteredRulesGeohashCounts = new Dictionary <string, Dictionary <PatternRule, Dictionary <string, int> > >();
var filteredRulesCounts = new Dictionary <string, Dictionary <PatternRule, int> >();
foreach (var domainToCounts in results.RulesGeohashCounts)
{
var domain = domainToCounts.Key;
var domainRulesToGeohashCounts = domainToCounts.Value;
var domainRulesCounts = results.RulesCounts[domain];
var filteredRulesToGeohashCounts = this.AddRetrieveDomainToRulesGeohashCounts(filteredRulesGeohashCounts, domain);
var filteredRulesToCounts = this.AddRetrieveDomainToRulesCounts(filteredRulesCounts, domain);
foreach (var ruleToGeohashCounts in domainRulesToGeohashCounts)
{
var rule = ruleToGeohashCounts.Key;
var geohashCounts = ruleToGeohashCounts.Value;
var ruleTotalCount = domainRulesCounts[rule];
foreach (var geohashCount in geohashCounts)
{
var geohash = geohashCount.Key;
var count = geohashCount.Value;
var occFraction = count / (1.0d * ruleTotalCount);
if (count >= minimumRuleSupport && occFraction >= minimimRuleOccFraction)
{
//Console.WriteLine($"{domain}\t{rule}\t{geohash}\t{count} / {ruleTotalCount} =\t{occFraction}");
filteredRulesToCounts[rule] = ruleTotalCount;
Dictionary <string, int> filteredGeohashCounts;
if (!filteredRulesToGeohashCounts.TryGetValue(rule, out filteredGeohashCounts))
{
filteredGeohashCounts = new Dictionary <string, int>();
filteredRulesToGeohashCounts[rule] = filteredGeohashCounts;
}
filteredGeohashCounts[geohash] = count;
}
}
}
}
return(new PatternMiningResult()
{
RulesGeohashCounts = filteredRulesGeohashCounts,
RulesCounts = filteredRulesCounts
});
}
public Dictionary <string, Dictionary <PatternRule, PatternMiningCoordinates> > ReduceRules(PatternMiningResult results)
{
var reducedRules = new Dictionary <string, Dictionary <PatternRule, PatternMiningCoordinates> >();
foreach (var domainToRulesEntry in results.RulesGeohashCounts)
{
var domain = domainToRulesEntry.Key;
var ruleCountsForDomain = results.RulesCounts[domain];
var rulesToGeohashCountsForDomain = domainToRulesEntry.Value;
Dictionary <PatternRule, PatternMiningCoordinates> rulesForDomain;
if (!reducedRules.TryGetValue(domain, out rulesForDomain))
{
rulesForDomain = new Dictionary <PatternRule, PatternMiningCoordinates>();
reducedRules[domain] = rulesForDomain;
}
foreach (var ruleEntry in rulesToGeohashCountsForDomain)
{
var rule = ruleEntry.Key;
var ruleTotalCount = ruleCountsForDomain[rule];
var geohashCounts = ruleEntry.Value;
rulesForDomain[rule] = FindBestCoordinates(geohashCounts, ruleTotalCount);
}
}
return(reducedRules);
}