Matrix Xchart 0.2.3

• add @CompileStatic to all 17 classes for performance and type safety (100% static compilation, no @CompileDynamic needed)
• complete empty test methods in HistogramChartTest (testDensityHistogram, testFrequencyHistogramCustom)
• add comprehensive GroovyDoc to all abstract classes (AbstractChart, AbstractXYChart, AbstractCategoryChart)
• add comprehensive edge case tests for improved test coverage
• replace Math calls with NumberExtension methods for idiomatic Groovy (Math.sqrt() → .sqrt())
• expand GroovyDoc on all public methods with detailed parameter descriptions and examples
• remove commented println debug statements
• fix IndexOutOfBoundsException in AbstractChart.makeFillTransparent() method
• fix multiple calculation issues in chart rendering
• update README version references and complete StickChart description

Matrix Tablesaw 0.2.2

Build Configuration Improvements

• Added compileTestJava configuration with deprecation and unchecked warnings enabled
• Added -Xlint:unchecked flag to compileGroovy for improved Groovy code quality checks
• Build configuration now consistent with matrix-core module standards

Code Quality Improvements

• Added @SuppressWarnings("unchecked") annotation to TableUtil.createColumn() method to properly handle intentional unchecked generic casts
• Removed duplicate BigDecimalColumn type check in classForColumnType() method (dead code removal)

Bug Fixes & Improvements

• BigDecimalColumn enhancements:
  • Fixed asBytes() method to use UTF-8 encoding explicitly instead of platform default charset, ensuring consistent byte representation across all platforms
  • Cleaned up asBytes() method documentation and removed outdated TODO comments
  • Extended toBigDecimal() method to handle additional Number subtypes:
    • BigDecimal - now returns the value as-is without conversion (prevents precision loss from unnecessary double conversion)
    • AtomicInteger - converted via get() for precision
    • AtomicLong - converted via get() for precision
    • DoubleAccumulator - converted via doubleValue()
  • Added comprehensive Javadoc explaining conversion behavior for all Number types
  • Improved test coverage to properly exercise toBigDecimal(Number) conversion path for BigDecimal inputs
  • Updated test assertions to use UTF-8 encoding for deterministic byte array comparisons

Documentation

• Enhanced Javadoc/Groovydoc documentation:
  • BigDecimalColumnFormatter - Added comprehensive class documentation with usage examples, documented all factory methods, constructors, and formatting methods
  • GdataFrameJoiner - Added class documentation explaining join types, documented all join method variants with parameter descriptions
  • Verified existing documentation in BigDecimalComparator, XlsxWriteOptions, and GdataFrameReader
• All public APIs now have production-quality documentation

Testing

• Added JaCoCo code coverage reporting infrastructure
  • Current coverage: 54% instruction coverage, 58% branch coverage
  • Coverage thresholds: 50% overall, 15% per class (baseline to prevent regression)
  • Coverage reports available in HTML and XML formats
  • Excluded low-coverage infrastructure classes from strict requirements
• Added test coverage for atomic type conversions in BigDecimalColumn
• Added test for BigDecimal precision preservation
• All 85 tests passing (2 new tests added)

Deprecations

• Deprecated OdsReadOptions.builder(Reader reader) - ODS is a binary format, not text-based
• Deprecated OdsReadOptions.builderFromString(String contents) - ODS is a binary format, not text-based
• Deprecated OdsReadOptions.builderFromUrl(String url) - ODS is a binary format, not text-based
• Note: These deprecated methods will be removed in v0.3.0

Dependency Updates

• com.github.miachm.sods:SODS [1.6.8 -> 1.7.0]
• org.apache.poi:poi-ooxml [5.4.1 -> 5.5.1]

Matrix Sql 2.3.0

• add option to control whether column names are quoted when creating a table
• add an execute method to MatrixSql to run arbitrary sql (update, delete, insert etc.)
• MatrixSqlFactory.create attempts to infer and set the JDBC driver when missing
• MatrixSql connection lifecycle fixes (reconnect after close)
• safer, prepared-statement updates with match-column validation
• ResultSet improvements: updateRow is a no-op for detached sets, null-safe primitive/stream getters, strict unwrap contract
• close metadata ResultSets for table discovery utilities
• Dependency upgrades:
  • commons-io:commons-io [2.20.0 -> 2.21.0]
  • se.alipsa.groovy:data-utils [2.0.3 -> 2.0.4]
  • se.alipsa:maven-3.9.11-utils [1.0.0 -> 1.1.0]

Matrix Spreadsheet 2.3.0

Major architectural refactoring with significant performance improvements

Breaking Changes

• removed POI and SODS implementations - FastExcel is now the single XLSX backend, FastOds is the single ODS backend
• removed ExcelImplementation and OdsImplementation enums - no implementation selection needed
• explicitly reject legacy .xls files (XLSX only)
• deprecated SpreadsheetExporter in favor of SpreadsheetWriter

New Features

• add append/replace support for existing XLSX and ODS files (preserves sheets and metadata)
• add flexible start position support when writing data (e.g., write to cell B5)
• add map-based multi-sheet API: writeSheets(Map<String, Position>)
• add new ODS streaming writer/appender with table attributes/column reuse for styling
• add profiling support for ODS operations via -Dmatrix.spreadsheet.ods.profile=true
• XLSX append now inherits sheetFormatPr, column widths, page margins, and fixes relId collisions
• add comprehensive sheet name sanitization with automatic de-duplication

Performance Improvements

• ODS read performance: 65-80% faster (medium files: 4.86s → 1.43s, large files: 262s → 53s)
• switched to Aalto StAX parser for 64% speedup
• adaptive row capacity sizing to minimize ArrayList resizing
• type-aware value extraction with switch dispatch
• optimized trailing empty row detection
• Null Object pattern for profiling eliminates branching overhead (~14% improvement)

Bug Fixes

• fix missing return statement in ValueExtractor.getDouble() (percentage parsing)
• fix 1-based sheet indexing consistency across all importers
• fix sheet name collision prevention (sanitization could cause silent data loss)
• fix invalid sheet number handling in URL imports
• fix null row guards in FExcelReader
• fix percentage parsing to be locale-independent
• fix race condition in SpreadsheetExporter static field sharing

Code Quality

• add column count validation to all write methods
• add robust cleanup for temp files and XML stream resources
• add XXE protection with hardened XML parsing
• replace 15+ println statements with proper logging
• remove ~500 lines of dead code (POI, SODS implementations)
• extract duplicate header building logic (DRY improvements)
• comprehensive test coverage: 79.74% (105 tests passing)
• add benchmarking suite for performance validation

Dependencies

• remove org.apache.poi:poi and org.apache.poi:poi-ooxml
• remove com.github.miachm.sods:SODS
• remove org.apache.logging.log4j:log4j-api (migrated to matrix-core Logger)
• add com.fasterxml:aalto-xml 1.3.4 (high-performance StAX parser)
• upgrade com.github.javaparser:javaparser-core [3.26.4 -> 3.27.0]
• migrate from log4j to matrix-core Logger (supports slf4j if present, otherwise System.out/err)

Matrix Smile 0.1.0

Initial release providing comprehensive integration between Matrix and Smile (Statistical Machine Intelligence and Learning Engine).

Core Features

• DataframeConverter: Bidirectional conversion between Matrix and Smile DataFrame with support for 18 data types
• SmileUtil: Pandas-like utility functions for data exploration and manipulation
  • Statistical summary (describe), column information (info), frequency tables
  • Sampling (random, by count, by fraction, with seed)
  • Head/tail operations, null detection and counting
• Gsmile Extension Module: Natural Groovy syntax extensions for Matrix and DataFrame
  • Matrix extensions: toSmileDataFrame(), smileDescribe(), smileSample()
  • DataFrame extensions: toMatrix(), subscript operators (getAt), filtering, iteration
  • Comprehensive test coverage (24 extension method tests)

Machine Learning Wrappers

• SmileClassifier: Wrappers for classification algorithms
  • Logistic Regression, Decision Trees, Random Forest, Gradient Boosted Trees
  • Support Vector Machines, K-Nearest Neighbors, Naive Bayes, AdaBoost
  • Model training, prediction, and evaluation with confusion matrices
• SmileRegression: Wrappers for regression algorithms
  • Linear Regression, Ridge Regression, LASSO, Elastic Net
  • Regression Trees, Gradient Boosted Trees, Random Forest
  • Model fitting, prediction, and RMSE calculation
• SmileCluster: Wrappers for clustering algorithms
  • K-Means, Hierarchical Clustering, DBSCAN, DENCLUE, CLARANS
  • Cluster assignment and centroids calculation
• SmileDimensionality: Dimensionality reduction techniques
  • PCA (Principal Component Analysis), MDS (Multidimensional Scaling)
  • t-SNE (t-Distributed Stochastic Neighbor Embedding)

Statistical Analysis (SmileStats)

• Probability Distributions:
  • Discrete: Binomial, Geometric, Poisson, Hypergeometric
  • Continuous: Normal, Exponential, Gamma, Beta, Chi-Squared, T, F, Weibull
  • PDF, CDF, quantile, and random sample generation
• Hypothesis Testing:
  • t-tests (one-sample, two-sample, paired)
  • Chi-squared test, F-test, Kolmogorov-Smirnov test
  • Correlation tests (Pearson, Spearman, Kendall) with significance testing
• Correlation Analysis:
  • Correlation matrices with p-values
  • Support for Pearson, Spearman, and Kendall correlation methods
• Random Sampling: Generate random samples from various distributions

Feature Engineering (SmileFeatures)

• Data Loading: Load datasets from Smile's built-in data repository
• Feature Scaling:
  • StandardScaler (z-score normalization with fit/transform workflow)
  • MinMaxScaler (range normalization)
  • MaxAbsScaler (maximum absolute value scaling)
  • RobustScaler (median and IQR-based scaling)
• Feature Encoding:
  • One-hot encoding for categorical variables
  • Label encoding for ordinal variables
• Feature Selection:
  • Sum, difference, product, ratio feature creation
• Imputation: Missing value handling with mean, median, mode, or constant strategies

Code Quality

• Comprehensive @CompileStatic annotation throughout for type safety and performance
• Modern Groovy 5.0+ switch expression syntax (arrow operators)
• Extensive GroovyDoc documentation (207 JavaDoc blocks)
• Comprehensive test coverage (274 tests across 10 test files, 100% test file coverage)
• Idiomatic Groovy code (as double instead of .doubleValue(), NumberExtension usage)

Dependencies

• com.github.haifengl:smile-core 4.4.2
• Requires Java 21 (Smile 4.x not compatible with Java 22+)
• Requires Groovy 5.0+ (for modern switch expression syntax)

Matrix Parquet 0.4.0

• remove parquet-carpet dependency (MatrixCarpetIO) - now using native Parquet implementation
• add support for nested structures: structs (POJOs, maps) and repeated fields (arrays)
• add URL, Path, InputStream, and byte[] input support to MatrixParquetReader (API consistency with matrix-csv and matrix-json)
• add BigDecimal precision and scale control in MatrixParquetWriter write methods
• add in-memory write support via InMemoryOutputFile and InMemoryPositionOutputStream (eliminates temporary files)
• add timezone support for timestamp handling (optional parameter in reader/writer methods)
• MatrixParquetWriter can now write to either a file or directory (using matrix name for filename)
• use matrixName as Parquet schema name if present
• add @CompileStatic to MatrixParquetReader for performance and type safety
• add comprehensive input validation to both reader and writer (null checks, empty matrix, file existence)
• add safeFileName sanitization for directory targets (strips path separators and unsafe characters)
• fix bug: time precision schema/implementation mismatch (now uses MICROS for timestamps, MILLIS for time)
• fix bug: BigDecimal schema inference incorrectly set minimum scale to 2
• extract magic strings to constants for maintainability
• add comprehensive GroovyDoc to MatrixParquetReader and MatrixParquetWriter
• add extensive test coverage including edge cases, validation, and round-trip verification
• cache reflection metadata for struct handling (performance optimization)
• upgrade dependencies
  • org.apache.hadoop:hadoop-common [3.4.1 -> 3.4.2]
  • org.apache.hadoop:hadoop-mapreduce-client-core [3.4.1 -> 3.4.2]
  • org.apache.parquet:parquet-column [1.15.2 -> 1.16.0]
  • org.apache.parquet:parquet-hadoop [1.15.2 -> 1.16.0]

Matrix Json 2.1.2

• deprecate JsonImporter and JsonExporter in favor of JsonReader and JsonWriter
• change implementation to use Jackson streaming API instead of JsonSlurper for improved memory efficiency (O(1) memory regardless of JSON size)
• add duplicate key detection in flatten() to prevent silent data loss
• add URL and Path support to JsonImporter (matching matrix-csv API)
• add static export methods to JsonExporter for API consistency
• add comprehensive test coverage for edge cases (empty arrays, single rows, null handling)
• add input validation to prevent writing empty or null matrices
• fix JsonImporter mutation and iteration assumptions
• fix TOCTOU race condition in JsonWriter file creation
• replace broad exception catches with specific exception types
• upgrade dependencies
  • com.fasterxml.jackson.core:jackson-core [2.20.0 -> 2.20.1]
  • com.fasterxml.jackson.core:jackson-databind [2.20.0 -> 2.20.1]

Matrix Gsheets 0.1.1

Move actual implementation for GsheetsReader and GsheetsWriter and utility methods to GsUtil so that GsImporter and GsExporter are just empty wrappers.

Matrix Groovy Ext 0.1.0

Initial version

• Number extensions allowing for more idiomatic groovy code.

Matrix csv 2.2.2

• deprecate CsvImporter and CsvExporter in favor of CsvReader and CsvWriter
• upgrade commons-csv from 1.14.0 to 1.14.1
• fix bug: empty CSV files now handled correctly (no more IndexOutOfBoundsException)
• fix typos in error messages ("extected" → "expected")
• add comprehensive test coverage for edge cases (empty CSV, header-only, single row/column, mismatched columns)
• add null validation to CsvExporter methods
• add GroovyDoc documentation to CsvImporter and CsvExporter classes and methods