public void GetAnovaStats(List <string> categories, List <string> numerals)
{
var anovaStats = new AnovaStats();
Dictionary <string, List <double> > groups = new Dictionary <string, List <double> >();
Dictionary <string, double> means = new Dictionary <string, double>();
for (var i = 0; i < categories.Count; i++)
{
//need to clean numbers...exclude $ , and stuff like that
if (categories[i] == "" || numerals[i] == "" || categories[i] == null || numerals[i] == null)
{
continue;
}
double numeral;
bool resultOfParse = double.TryParse(numerals[i], out numeral);
if (resultOfParse)
{
if (!groups.ContainsKey(categories[i]))
{
groups.Add(categories[i], new List <double> {
numeral
});
}
else
{
groups[categories[i]].Add(numeral);
}
}
}
double ssb = 0;
double ssw = 0;
double sst = 0;
int dfB;
int dfW;
double f;
//df bt is #kvp -1
//dfw
double indTotalSquare = 0;
double totalSum = 0;
int totalCount = 0;
foreach (KeyValuePair <string, List <double> > kvp in groups)
{
var mean = kvp.Value.Sum() / kvp.Value.Count();
means.Add(kvp.Key, mean);
//for ssb
indTotalSquare += (Math.Pow(kvp.Value.Sum(), 2)) / kvp.Value.Count();
totalSum += kvp.Value.Sum();
totalCount += kvp.Value.Count();
//for ssw
foreach (double num in kvp.Value)
{
ssw += (Math.Pow(num - mean, 2));
}
}
ssb = indTotalSquare - (Math.Pow(totalSum, 2) / totalCount);
sst = ssb + ssw;
dfB = groups.Count - 1;
dfW = totalCount - groups.Count;
f = (ssb / (double)dfB) / (ssw / (double)dfW);
//Console.WriteLine("ssb: {0}, sst: {1}, dfB: {2}, dfW: {3}, ssw: {4}, f: {5}", ssb, sst, dfB, dfW, ssw, f);
var keyValueMeans = new List <KeyValue>();
foreach (KeyValuePair <string, double> kvp in means)
{
keyValueMeans.Add(new KeyValue {
Key = kvp.Key, Value = kvp.Value
});
}
FTable fTable = new FTable();
CriticalValueAtAlphaZeroFive = fTable.sigAtZeroFive(dfB, dfW, f);
SigAtPointZeroFive = f > CriticalValueAtAlphaZeroFive;
SignificantResult = f > CriticalValueAtAlphaZeroFive;
Means = keyValueMeans;
Groups = groups;
FStat = f;
}
public static StatSummaryVM GetSummaryVM(Dictionary <string, List <string> > dictFile, SelectTypeVM vm)
{
ApplicationDbContext _context = new ApplicationDbContext();
List <string> checkedColumns = new List <string>();
List <ChiStats> chiStatsList = new List <ChiStats>();
List <AnovaStats> anovaStatsList = new List <AnovaStats>();
List <PearsonCorr> pearsonCorrelations = new List <PearsonCorr>();
List <DateAndCategory> dateAndCategories = new List <DateAndCategory>();
List <DateAndNumeral> dateAndNumerals = new List <DateAndNumeral>();
StatSummaryVM statsSummary = new StatSummaryVM();
foreach (KeyValuePair <string, List <string> > entry in dictFile)
{
foreach (KeyValuePair <string, List <string> > entryCompare in dictFile)
{
var checkedKey = Methods.ConcatAlph(entry.Key, entryCompare.Key);
if (entry.Key == entryCompare.Key ||
checkedColumns.Contains(checkedKey) ||
(entry.Value.Count != entryCompare.Value.Count) ||
(vm.ColumnTypes[entry.Key] == "exclude" || vm.ColumnTypes[entryCompare.Key] == "exclude"))
{
continue;
}
if ((vm.ColumnTypes[entry.Key] == "date time" || vm.ColumnTypes[entryCompare.Key] == "date time")
&&
(vm.ColumnTypes[entry.Key] == "categorical" || vm.ColumnTypes[entryCompare.Key] == "categorical"))
{
DateAndCategory dateCat = new DateAndCategory();
if (vm.ColumnTypes[entry.Key] == "date time")
{
dateCat.GetLinePlotData(entry.Value, entryCompare.Value);
}
else
{
dateCat.GetLinePlotData(entryCompare.Value, entry.Value);
}
dateCat.Variable1 = entry.Key;
dateCat.Variable2 = entryCompare.Key;
dateAndCategories.Add(dateCat);
//_context.SaveChanges();
}
if ((vm.ColumnTypes[entry.Key] == "date time" || vm.ColumnTypes[entryCompare.Key] == "date time")
&&
(vm.ColumnTypes[entry.Key] == "numeral" || vm.ColumnTypes[entryCompare.Key] == "numeral"))
{
DateAndNumeral dateNum = new DateAndNumeral();
if (vm.ColumnTypes[entry.Key] == "date time")
{
dateNum.MakeDataBlob(entry.Value, entryCompare.Value);
dateNum.DateName = entry.Key;
dateNum.NumeralName = entryCompare.Key;
}
else
{
dateNum.MakeDataBlob(entryCompare.Value, entry.Value);
dateNum.DateName = entryCompare.Key;
dateNum.NumeralName = entry.Key;
}
dateAndNumerals.Add(dateNum);
//_context.SaveChanges();
}
if (vm.ColumnTypes[entry.Key] == "categorical" && vm.ColumnTypes[entryCompare.Key] == "categorical")
{
ChiStats chiStats = new ChiStats();
chiStats.GetChiStat(entry.Value, entryCompare.Value);
chiStats.Variable1 = entryCompare.Key.Replace("\"", "");
chiStats.Variable2 = entry.Key.Replace("\"", "");
chiStatsList.Add(chiStats);
//_context.ChiStats.Add(chiStats);
//_context.SaveChanges();
}
if ((vm.ColumnTypes[entry.Key] == "categorical" || vm.ColumnTypes[entryCompare.Key] == "categorical")
&&
(vm.ColumnTypes[entry.Key] == "numeral" || vm.ColumnTypes[entryCompare.Key] == "numeral"))
{
AnovaStats anovaStats = new AnovaStats();
if (vm.ColumnTypes[entry.Key] == "numeral")
{
anovaStats.GetAnovaStats(entryCompare.Value, entry.Value);
anovaStats.CategoricalVariable = entryCompare.Key;
anovaStats.NumericalVariable = entry.Key;
anovaStatsList.Add(anovaStats);
}
else
{
anovaStats.GetAnovaStats(entry.Value, entryCompare.Value);
anovaStats.CategoricalVariable = entry.Key;
anovaStats.NumericalVariable = entryCompare.Key;
anovaStatsList.Add(anovaStats);
}
//anovaStats.SaveStat();
_context.AnovaStats.Add(anovaStats);
_context.SaveChanges();
}
if (vm.ColumnTypes[entry.Key] == "numeral" && vm.ColumnTypes[entryCompare.Key] == "numeral")
{
PearsonCorr pearsonCorr = new PearsonCorr();
pearsonCorr.Variable1 = entry.Key;
pearsonCorr.Variable2 = entryCompare.Key;
pearsonCorr.ComputeCoeff(entry.Value, entryCompare.Value);
pearsonCorrelations.Add(pearsonCorr);
_context.SaveChanges();
}
checkedColumns.Add(checkedKey);
}
}
statsSummary.Name = vm.Name != null ? vm.Name : "Untitled";
statsSummary.Description = vm.Description != null ? vm.Description : "no description available";
statsSummary.AnovaStats = anovaStatsList;
statsSummary.ChiStats = chiStatsList;
statsSummary.PearsonCorrs = pearsonCorrelations;
statsSummary.DateAndCatories = dateAndCategories;
statsSummary.DateAndNumerals = dateAndNumerals;
statsSummary.Path = vm.Path;
statsSummary.FileName = vm.FileName;
return(statsSummary);
}
