C++ (Cpp) CountMinSketchAdd Examples

Example #1

File: test_cmsketch.c Project: jberkus/pipelinedb

END_TEST

START_TEST(test_union)
{
	CountMinSketch *cms1 = CountMinSketchCreateWithEpsAndP(EPS, P);
	CountMinSketch *cms2 = CountMinSketchCreateWithEpsAndP(EPS, P);
	CountMinSketch *cms;
	int num_keys = 5000;
	int *keys = palloc(sizeof(int) * num_keys * 2);
	int *counts = palloc0(sizeof(int) * num_keys * 2);
	int i;

	for (i = 0; i < num_keys; i++)
	{
		keys[i] = rand();
		keys[num_keys + i] = rand();
	}

	for (i = 0; i < num_keys; i++)
	{
		counts[i] += rand() % 100;
		counts[num_keys + i] = rand() % 100;
		CountMinSketchAdd(cms1, &keys[i], sizeof(int), counts[i]);
		CountMinSketchAdd(cms2, &keys[num_keys + i], sizeof(int), counts[num_keys + i]);
	}

	cms = CountMinSketchMerge(cms1, cms2);

	for (i = 0; i < num_keys * 2; i++)
	{
		float diff = CountMinSketchEstimateFrequency(cms, &keys[i], sizeof(int)) - counts[i];
		ck_assert(diff / cms->count < EPS);
	}
}

Example #2

File: test_cmsketch.c Project: jberkus/pipelinedb

END_TEST

START_TEST(test_false_positives)
{
	CountMinSketch *cms = CountMinSketchCreateWithEpsAndP(EPS, P);
	int range = 1 << 20;
	int num_keys = 1 << 16;
	int *counts = palloc0(sizeof(int) * range);
	int *keys = palloc(sizeof(int) * num_keys);
	int i;
	float num_errors = 0;

	for (i = 0; i < num_keys; i++)
		keys[i] = rand() % range;

	for (i = 0; i < num_keys; i++)
	{
		int c = rand() % 100;
		counts[keys[i]] += c;
		CountMinSketchAdd(cms, &keys[i], sizeof(int), c);
	}

	for (i = 0; i < range; i++)
	{
		float diff = 1.0 * abs(CountMinSketchEstimateFrequency(cms, &i, sizeof(int)) - counts[i]);
		if (diff / cms->count > EPS)
			num_errors++;
	}

	pfree(counts);
	pfree(keys);

	ck_assert(num_errors / range < (1 - P));
}

Example #3

File: cmsketchfuncs.c Project: seco/pipelinedb

static CountMinSketch *
cmsketch_add_datum(FunctionCallInfo fcinfo, CountMinSketch *cms, Datum elem)
{
	TypeCacheEntry *typ = (TypeCacheEntry *) fcinfo->flinfo->fn_extra;
	Size size;

	if (!typ->typbyval && !elem)
		return cms;

	size = datumGetSize(elem, typ->typbyval, typ->typlen);

	if (typ->typbyval)
		CountMinSketchAdd(cms, (char *) &elem, size, 1);
	else
		CountMinSketchAdd(cms, DatumGetPointer(elem), size, 1);

	return cms;
}

Example #4

File: cmsketchfuncs.c Project: jberkus/pipelinedb

static CountMinSketch *
cmsketch_add_datum(FunctionCallInfo fcinfo, CountMinSketch *cms, Datum elem, int32 n)
{
	TypeCacheEntry *typ = (TypeCacheEntry *) fcinfo->flinfo->fn_extra;
	StringInfo buf;

	if (!typ->typbyval && !elem)
		return cms;

	buf = makeStringInfo();
	DatumToBytes(elem, typ, buf);
	CountMinSketchAdd(cms, buf->data, buf->len, n);

	pfree(buf->data);
	pfree(buf);

	return cms;
}