Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix!: statistical functions should return null when provided a vector of only null values #5606

Merged
merged 19 commits into from
Jun 19, 2024
Merged

fix!: statistical functions should return null when provided a vector of only null values #5606

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
19 commits
Select commit Hold shift + click to select a range
076e286
Updated Numeric calculations and tests, corrected median and percenti…
lbooker42 Jun 11, 2024
dd9f507
Correct Numeric avg() function and tests.
lbooker42 Jun 12, 2024
5bf465e
UpdateBy corrections for all-NULL vectors
lbooker42 Jun 12, 2024
ad2675d
Agg corrections for avg,var,std
lbooker42 Jun 12, 2024
50667f2
Agg test changes to reflect new output.
lbooker42 Jun 13, 2024
ff6f010
Spotless
lbooker42 Jun 13, 2024
9ff3cf6
Tiny change for correctness.
lbooker42 Jun 13, 2024
8753267
Addressed PR comments.
lbooker42 Jun 17, 2024
4481f79
Addressed another round of PR comments.
lbooker42 Jun 17, 2024
9422488
More comments addressed.
lbooker42 Jun 17, 2024
5750546
Bug fix to pass OOB.
lbooker42 Jun 18, 2024
b06a063
Merge branch 'main' into lab-numeric-harmony
lbooker42 Jun 18, 2024
35c03d1
Spotless
lbooker42 Jun 18, 2024
0f150ac
Undo test commenting.
lbooker42 Jun 18, 2024
4e01d9d
Better testing and coverage, fix to wvar() and misc cleanup.
lbooker42 Jun 18, 2024
2ea6418
Test correction.
lbooker42 Jun 18, 2024
e00e49a
Removed dead code.
lbooker42 Jun 18, 2024
6bf8bb9
Update engine/function/src/templates/Numeric.ftl
lbooker42 Jun 18, 2024
cf1f22e
Update engine/function/src/templates/Numeric.ftl
lbooker42 Jun 18, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 13 additions & 11 deletions engine/table/src/main/java/io/deephaven/engine/table/impl/by/ByteChunkedVarOperator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,7 @@
// @formatter:off
package io.deephaven.engine.table.impl.by;

import io.deephaven.base.verify.Assert;
import io.deephaven.chunk.attributes.ChunkLengths;
import io.deephaven.chunk.attributes.ChunkPositions;
import io.deephaven.chunk.attributes.Values;
Expand Down Expand Up @@ -87,24 +88,25 @@ private boolean addChunk(ByteChunk<? extends Values> values, long destination, i
final double sum = SumByteChunk.sum2ByteChunk(values, chunkStart, chunkSize, chunkNonNull, sum2);

if (chunkNonNull.get() > 0) {
final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, chunkNonNull.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, chunkNonNull.get());
final double newSum = plusDouble(sumSource.getUnsafe(destination), sum);
final double newSum2 = plusDouble(sum2Source.getUnsafe(destination), sum2.doubleValue());

sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

if (nonNullCount <= 1) {
Assert.neqZero(totalNormalCount, "totalNormalCount");
if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
} else {
final double variance = (newSum2 - (newSum * newSum / nonNullCount)) / (nonNullCount - 1);
final double variance = (newSum2 - (newSum * newSum / totalNormalCount)) / (totalNormalCount - 1);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);
}
} else {
final long nonNullCount = nonNullCounter.getCountUnsafe(destination);
if (nonNullCount == 0) {
final long totalNormalCount = nonNullCounter.getCountUnsafe(destination);
if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
} else if (nonNullCount == 1) {
} else if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
}
}
Expand All @@ -120,12 +122,12 @@ private boolean removeChunk(ByteChunk<? extends Values> values, long destination
return false;
}

final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, -chunkNonNull.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, -chunkNonNull.get());

final double newSum;
final double newSum2;

if (nonNullCount == 0) {
if (totalNormalCount == 0) {
newSum = newSum2 = 0;
} else {
newSum = plusDouble(sumSource.getUnsafe(destination), -sum);
Expand All @@ -135,15 +137,15 @@ private boolean removeChunk(ByteChunk<? extends Values> values, long destination
sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

if (nonNullCount == 0) {
if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
return true;
} else if (nonNullCount == 1) {
} else if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
return true;
}

final double variance = (newSum2 - (newSum * newSum / nonNullCount)) / (nonNullCount - 1);
final double variance = (newSum2 - (newSum * newSum / totalNormalCount)) / (totalNormalCount - 1);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);

return true;
Expand Down
24 changes: 13 additions & 11 deletions engine/table/src/main/java/io/deephaven/engine/table/impl/by/CharChunkedVarOperator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
//
package io.deephaven.engine.table.impl.by;

import io.deephaven.base.verify.Assert;
import io.deephaven.chunk.attributes.ChunkLengths;
import io.deephaven.chunk.attributes.ChunkPositions;
import io.deephaven.chunk.attributes.Values;
Expand Down Expand Up @@ -83,24 +84,25 @@ private boolean addChunk(CharChunk<? extends Values> values, long destination, i
final double sum = SumCharChunk.sum2CharChunk(values, chunkStart, chunkSize, chunkNonNull, sum2);

if (chunkNonNull.get() > 0) {
final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, chunkNonNull.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, chunkNonNull.get());
final double newSum = plusDouble(sumSource.getUnsafe(destination), sum);
final double newSum2 = plusDouble(sum2Source.getUnsafe(destination), sum2.doubleValue());

sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

if (nonNullCount <= 1) {
Assert.neqZero(totalNormalCount, "totalNormalCount");
if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
} else {
final double variance = (newSum2 - (newSum * newSum / nonNullCount)) / (nonNullCount - 1);
final double variance = (newSum2 - (newSum * newSum / totalNormalCount)) / (totalNormalCount - 1);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);
}
} else {
final long nonNullCount = nonNullCounter.getCountUnsafe(destination);
if (nonNullCount == 0) {
final long totalNormalCount = nonNullCounter.getCountUnsafe(destination);
if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
} else if (nonNullCount == 1) {
} else if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
}
}
Expand All @@ -116,12 +118,12 @@ private boolean removeChunk(CharChunk<? extends Values> values, long destination
return false;
}

final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, -chunkNonNull.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, -chunkNonNull.get());

final double newSum;
final double newSum2;

if (nonNullCount == 0) {
if (totalNormalCount == 0) {
newSum = newSum2 = 0;
} else {
newSum = plusDouble(sumSource.getUnsafe(destination), -sum);
Expand All @@ -131,15 +133,15 @@ private boolean removeChunk(CharChunk<? extends Values> values, long destination
sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

if (nonNullCount == 0) {
if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
return true;
} else if (nonNullCount == 1) {
} else if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
return true;
}

final double variance = (newSum2 - (newSum * newSum / nonNullCount)) / (nonNullCount - 1);
final double variance = (newSum2 - (newSum * newSum / totalNormalCount)) / (totalNormalCount - 1);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);

return true;
Expand Down
19 changes: 11 additions & 8 deletions engine/table/src/main/java/io/deephaven/engine/table/impl/by/DoubleChunkedVarOperator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,23 +97,23 @@ private boolean addChunk(DoubleChunk<? extends Values> values, long destination,
final boolean forceNanResult = totalNegativeInfinities > 0 || totalPositiveInfinities > 0 || totalNanCount > 0;

if (chunkNormalCount.get() > 0) {
final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, chunkNormalCount.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, chunkNormalCount.get());
final double newSum = NullSafeAddition.plusDouble(sumSource.getUnsafe(destination), sum);
final double newSum2 = NullSafeAddition.plusDouble(sum2Source.getUnsafe(destination), sum2.doubleValue());

sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

Assert.neqZero(nonNullCount, "nonNullCount");
if (forceNanResult || nonNullCount == 1) {
Assert.neqZero(totalNormalCount, "totalNormalCount");
if (forceNanResult || totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
} else {
// If the sum or sumSquared has reached +/-Infinity, we are stuck with NaN forever.
if (Double.isInfinite(newSum) || Double.isInfinite(newSum2)) {
resultColumn.set(destination, Double.NaN);
return true;
}
final double variance = computeVariance(nonNullCount, newSum, newSum2);
final double variance = computeVariance(totalNormalCount, newSum, newSum2);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);
}
return true;
Expand Down Expand Up @@ -176,15 +176,17 @@ private boolean removeChunk(DoubleChunk<? extends Values> values, long destinati
}
sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);
} else if (totalNormalCount == 1 || forceNanResult) {
} else {
newSum = sumSource.getUnsafe(destination);
newSum2 = sum2Source.getUnsafe(destination);
}

if (totalNormalCount == 1 || forceNanResult) {
resultColumn.set(destination, Double.NaN);
return true;
} else if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
return true;
} else {
newSum = sumSource.getUnsafe(destination);
newSum2 = sum2Source.getUnsafe(destination);
}

// If the sum has reach +/-Infinity, we are stuck with NaN forever.
Expand All @@ -196,6 +198,7 @@ private boolean removeChunk(DoubleChunk<? extends Values> values, long destinati
// Perform the calculation in a way that minimizes the impact of FP error.
final double variance = computeVariance(totalNormalCount, newSum, newSum2);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);

return true;
}

Expand Down
19 changes: 11 additions & 8 deletions engine/table/src/main/java/io/deephaven/engine/table/impl/by/FloatChunkedVarOperator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,23 +93,23 @@ private boolean addChunk(FloatChunk<? extends Values> values, long destination,
final boolean forceNanResult = totalNegativeInfinities > 0 || totalPositiveInfinities > 0 || totalNanCount > 0;

if (chunkNormalCount.get() > 0) {
final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, chunkNormalCount.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, chunkNormalCount.get());
final double newSum = NullSafeAddition.plusDouble(sumSource.getUnsafe(destination), sum);
final double newSum2 = NullSafeAddition.plusDouble(sum2Source.getUnsafe(destination), sum2.doubleValue());

sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

Assert.neqZero(nonNullCount, "nonNullCount");
if (forceNanResult || nonNullCount == 1) {
Assert.neqZero(totalNormalCount, "totalNormalCount");
if (forceNanResult || totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
} else {
// If the sum or sumSquared has reached +/-Infinity, we are stuck with NaN forever.
if (Double.isInfinite(newSum) || Double.isInfinite(newSum2)) {
resultColumn.set(destination, Double.NaN);
return true;
}
final double variance = computeVariance(nonNullCount, newSum, newSum2);
final double variance = computeVariance(totalNormalCount, newSum, newSum2);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);
}
return true;
Expand Down Expand Up @@ -172,15 +172,17 @@ private boolean removeChunk(FloatChunk<? extends Values> values, long destinatio
}
sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);
} else if (totalNormalCount == 1 || forceNanResult) {
} else {
newSum = sumSource.getUnsafe(destination);
newSum2 = sum2Source.getUnsafe(destination);
}

if (totalNormalCount == 1 || forceNanResult) {
resultColumn.set(destination, Double.NaN);
return true;
} else if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
return true;
} else {
newSum = sumSource.getUnsafe(destination);
newSum2 = sum2Source.getUnsafe(destination);
}

// If the sum has reach +/-Infinity, we are stuck with NaN forever.
Expand All @@ -192,6 +194,7 @@ private boolean removeChunk(FloatChunk<? extends Values> values, long destinatio
// Perform the calculation in a way that minimizes the impact of FP error.
final double variance = computeVariance(totalNormalCount, newSum, newSum2);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);

return true;
}

Expand Down
24 changes: 13 additions & 11 deletions engine/table/src/main/java/io/deephaven/engine/table/impl/by/IntChunkedVarOperator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,7 @@
// @formatter:off
package io.deephaven.engine.table.impl.by;

import io.deephaven.base.verify.Assert;
import io.deephaven.chunk.attributes.ChunkLengths;
import io.deephaven.chunk.attributes.ChunkPositions;
import io.deephaven.chunk.attributes.Values;
Expand Down Expand Up @@ -87,24 +88,25 @@ private boolean addChunk(IntChunk<? extends Values> values, long destination, in
final double sum = SumIntChunk.sum2IntChunk(values, chunkStart, chunkSize, chunkNonNull, sum2);

if (chunkNonNull.get() > 0) {
final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, chunkNonNull.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, chunkNonNull.get());
final double newSum = plusDouble(sumSource.getUnsafe(destination), sum);
final double newSum2 = plusDouble(sum2Source.getUnsafe(destination), sum2.doubleValue());

sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

if (nonNullCount <= 1) {
Assert.neqZero(totalNormalCount, "totalNormalCount");
if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
} else {
final double variance = (newSum2 - (newSum * newSum / nonNullCount)) / (nonNullCount - 1);
final double variance = (newSum2 - (newSum * newSum / totalNormalCount)) / (totalNormalCount - 1);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);
}
} else {
final long nonNullCount = nonNullCounter.getCountUnsafe(destination);
if (nonNullCount == 0) {
final long totalNormalCount = nonNullCounter.getCountUnsafe(destination);
if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
} else if (nonNullCount == 1) {
} else if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
}
}
Expand All @@ -120,12 +122,12 @@ private boolean removeChunk(IntChunk<? extends Values> values, long destination,
return false;
}

final long nonNullCount = nonNullCounter.addNonNullUnsafe(destination, -chunkNonNull.get());
final long totalNormalCount = nonNullCounter.addNonNullUnsafe(destination, -chunkNonNull.get());

final double newSum;
final double newSum2;

if (nonNullCount == 0) {
if (totalNormalCount == 0) {
newSum = newSum2 = 0;
} else {
newSum = plusDouble(sumSource.getUnsafe(destination), -sum);
Expand All @@ -135,15 +137,15 @@ private boolean removeChunk(IntChunk<? extends Values> values, long destination,
sumSource.set(destination, newSum);
sum2Source.set(destination, newSum2);

if (nonNullCount == 0) {
if (totalNormalCount == 0) {
resultColumn.set(destination, NULL_DOUBLE);
return true;
} else if (nonNullCount == 1) {
} else if (totalNormalCount == 1) {
resultColumn.set(destination, Double.NaN);
return true;
}

final double variance = (newSum2 - (newSum * newSum / nonNullCount)) / (nonNullCount - 1);
final double variance = (newSum2 - (newSum * newSum / totalNormalCount)) / (totalNormalCount - 1);
resultColumn.set(destination, std ? Math.sqrt(variance) : variance);

return true;
Expand Down
Loading