protected async override Task <Result> Explore()
{
var stats = await statsResultProvider.ResultAsync;
var bucketsToSample = BucketUtils.EstimateBucketResolutions(
stats.Count,
stats.Min,
stats.Max,
ValuesPerBucketTarget,
isIntegerColumn: ctx.ColumnType == DValueType.Integer);
var histogramQ = await conn.Exec(new SingleColumnHistogram(ctx.Table, ctx.Column, bucketsToSample));
var histograms = Histogram.FromQueryRows(histogramQ.Rows);
var valueCounts = histogramQ.Rows
.GroupBy(
row => row.BucketSize,
(bs, rows) => (BucketSize: new BucketSize(bs), Rows: ValueCounts.Compute(rows)));
var results = valueCounts.Join(
histograms,
v => v.BucketSize.SnappedSize,
h => h.BucketSize.SnappedSize,
(v, h) => new Result(v.Rows, h));
return(results
.OrderBy(h => h.BucketSize.SnappedSize)
.ThenBy(h => h.ValueCounts.SuppressedCount)
.First());
}
public void BucketRange(int index, int bucketRange, string expected)
{
//arrange
//act
String result = BucketUtils.BucketRange(index, bucketRange);
//assert
Assert.Equal(expected, result);
}
protected async override Task <List <HistogramWithCounts>?> Explore()
{
var stats = await statsResultProvider.ResultAsync;
if (stats == null)
{
return(null);
}
var(minBound, maxBound) = (stats.Min, stats.Max);
if (!minBound.HasValue || !maxBound.HasValue)
{
var distincts = await distinctValuesProvider.ResultAsync;
if (distincts == null || distincts.ValueCounts.NonSuppressedNonNullCount == 0)
{
return(null);
}
var values = distincts.DistinctRows.Where(row => row.HasValue).Select(row => row.Value.GetDouble());
minBound ??= values.Min();
maxBound ??= values.Max();
}
if (!minBound.HasValue || !maxBound.HasValue || minBound == maxBound)
{
Logger.LogWarning("Unable to calculate suitable bounds for numerical column {Context.Column}.");
return(null);
}
var bucketsToSample = BucketUtils.EstimateBucketResolutions(
stats.Count,
(double)minBound,
(double)maxBound,
ValuesPerBucketTarget,
isIntegerColumn: Context.ColumnInfo.Type == DValueType.Integer);
var histogramQ = await Context.Exec(new SingleColumnHistogram(bucketsToSample));
var histograms = Histogram.FromQueryRows(histogramQ.Rows);
var valueCounts = histogramQ.Rows
.GroupBy(
row => row.BucketSize,
(bs, rows) => (BucketSize: new BucketSize(bs), Rows: ValueCounts.Compute(rows)));
return(valueCounts
.Join(
histograms,
v => v.BucketSize.SnappedSize,
h => h.BucketSize.SnappedSize,
(v, h) => new HistogramWithCounts(v.Rows, h))
.ToList());
}
private List <string> GenerateHeaders(IEnumerable <CohortGroup> cohortGroups)
{
List <string> headers = new List <string> {
"Cohort",
"Customers",
};
var maximumBuckets = cohortGroups
.Max(item => item.Buckets.Count());
for (int i = 0; i < maximumBuckets; i++)
{
headers
.Add($"{BucketUtils.BucketRange(i, _settings.LifeCycleRange)} days");
}
return(headers);
}
