[Autoloop] [Autoloop: build-tsb-pandas-typescript-migration] Iteration 72: value_counts + elem_ops exports#57
Merged
mrjf merged 20 commits intoApr 7, 2026
Conversation
Port pandas.core.groupby — DataFrameGroupBy and SeriesGroupBy with full split-apply-combine engine: - DataFrameGroupBy: sum, mean, min, max, count, std, first, last, size, agg (named/fn/per-column spec), transform, apply, filter, getGroup, ngroups, groupKeys, groups — single-key and multi-key support - SeriesGroupBy: same aggregation API, transform, apply, filter, getGroup - DataFrame.groupby(by) and Series.groupby(by) convenience methods - 40+ unit tests + property-based tests (fast-check) - Interactive playground page (playground/groupby.html) Run: https://github.com/githubnext/tsessebe/actions/runs/23972003902 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Implement pandas.concat() — combine Series and DataFrames along either axis. ## What's included **src/merge/concat.ts** — full concat implementation: - axis=0 (default): stack Series → Series; stack DataFrames → DataFrame - axis=1: Series[] → DataFrame (each Series as a column); DataFrame[] → DataFrame (side-by-side) - join='outer' (default): union of labels, null-fill missing values - join='inner': intersection of labels only - ignoreIndex: reset the result to a RangeIndex **src/merge/index.ts** — barrel export **src/index.ts** — exports concat + ConcatOptions **tests/merge/concat.test.ts** — 25+ unit tests + 4 property-based tests (fast-check) **playground/concat.html** — interactive tutorial page with 7 examples **playground/index.html** — concat marked as ✅ Complete Run: https://github.com/githubnext/tsessebe/actions/runs/23972580333 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Port pandas.merge — SQL-style join of two DataFrames. - src/merge/merge.ts: full merge() implementation with inner/left/right/outer joins, on/left_on/right_on/left_index/right_index, suffix handling, sort - tests/merge/merge.test.ts: 29 tests (unit + property-based with fast-check) - playground/merge.html: interactive tutorial with all join types - playground/index.html: mark merge as complete Metric: pandas_features_ported 8 → 9 Run: https://github.com/githubnext/tsessebe/actions/runs/24001239424 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Port pandas StringMethods as Series.str accessor. - src/core/string_accessor.ts: StringAccessor class with 35+ methods: case (lower/upper/title/capitalize/swapcase), len, strip/lstrip/rstrip, pad/ljust/rjust/center/zfill, contains/startswith/endswith/match/fullmatch, find/rfind/count, replace/extract, split/rsplit/join/cat, slice/get/sliceReplace, repeat/wrap, encode, isalpha/isdigit/isalnum/islower/isupper/istitle/isspace. All methods propagate null/NaN unchanged. - src/core/series.ts: added withValues() helper + Series.str getter - tests/core/string_accessor.test.ts: 60 tests (unit + property-based with fast-check) - playground/string_accessor.html: interactive tutorial with 9 sections - playground/index.html: mark str accessor as complete Metric: pandas_features_ported 9 → 10 Run: https://github.com/githubnext/tsessebe/actions/runs/24001823414 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Port pandas DatetimeProperties as Series.dt accessor.
- src/core/datetime_accessor.ts: DatetimeAccessor class with:
Calendar components: year/month/day/hour/minute/second/millisecond/
microsecond/nanosecond/dayofweek/weekday/dayofyear/quarter/
isocalendar_week/days_in_month/daysinmonth
Boolean boundaries: is_month_start/is_month_end/is_quarter_start/
is_quarter_end/is_year_start/is_year_end/is_leap_year
Formatting: strftime() with 25+ directives (%Y/%m/%d/%H/%M/%S/%A/%B etc.)
Normalization: normalize() (floor to midnight), date()
Rounding: floor/ceil/round with D/H/T/min/S/L/ms units
Epoch: total_seconds()
All methods propagate null/undefined/NaN unchanged
- src/core/series.ts: added Series.dt getter returning DatetimeAccessor
- src/core/index.ts: export DatetimeAccessor + DatetimeSeriesLike
- src/index.ts: re-export DatetimeAccessor + DatetimeSeriesLike
- tests/core/datetime_accessor.test.ts: 50+ tests (unit + property-based with fast-check)
- playground/datetime_accessor.html: interactive tutorial with 8 sections
- playground/index.html: mark dt accessor as complete
Metric: pandas_features_ported 10 → 11
Run: https://github.com/githubnext/tsessebe/actions/runs/24002454105
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Implements src/stats/describe.ts — the pandas-style describe() function: - quantile(sorted, q): linear interpolation (pandas method='linear') - describe(series): count/mean/std/min/percentiles/max for numeric, count/unique/top/freq for categorical - describe(dataFrame): per-column stats, include='number'|'object'|'all' - DescribeOptions: custom percentiles array - Series.quantile(q) method added to Series New files: src/stats/describe.ts, src/stats/index.ts Tests: tests/stats/describe.test.ts (32 tests, unit + property-based) Playground: playground/describe.html with 7 interactive tutorials Metric: pandas_features_ported 11 → 12 Run: https://github.com/githubnext/tsessebe/actions/runs/24002845454 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add src/io/csv.ts — pandas-style CSV read/write: - readCsv(text, options?): parse CSV string → DataFrame with automatic dtype inference (bool/int64/float64/string/object), NA handling, quoted fields (RFC 4180), custom separators, indexCol, skipRows, nRows - toCsv(df, options?): serialize DataFrame → CSV string with header, index, custom sep, lineterminator, naRep - Comprehensive tests: 35+ unit + property-based (fast-check) tests - Interactive playground: playground/csv.html with 7 tutorial sections Metric: 12 → 13 (+1) Run: https://github.com/githubnext/tsessebe/actions/runs/24003267099 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Five orient formats (records, split, index, columns, values) mirroring pandas read_json() and DataFrame.to_json(). Full null propagation, dtype override support, and JSON auto-detection. 31 unit + property-based tests. Interactive playground: playground/json.html. Run: https://github.com/githubnext/tsessebe/actions/runs/24003815679 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add `src/stats/corr.ts` with: - `pearsonCorr(a, b, options?)` — Pearson correlation for two Series, aligning on shared index labels and dropping missing pairs - `dataFrameCorr(df, options?)` — symmetric N×N Pearson correlation matrix for all numeric columns of a DataFrame - `dataFrameCov(df, options?)` — symmetric N×N sample covariance matrix Also adds inline methods: - `Series.corr(other, minPeriods?)` - `DataFrame.corr(minPeriods?)` - `DataFrame.cov(ddof?, minPeriods?)` 34 unit + property-based tests. Playground: playground/corr.html. Run: https://github.com/githubnext/tsessebe/actions/runs/24004259683 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
… DataFrame.rolling) - Add `src/window/rolling.ts` — `Rolling` class with full pandas-compatible sliding-window API: mean(), sum(), std(), var(), min(), max(), count(), median(), apply(). Uses `RollingSeriesLike` interface to avoid circular imports. Supports `minPeriods` and `center` options. - Add `DataFrameRolling` to `src/core/frame.ts` — column-wise rolling aggregations for DataFrame (same methods as Rolling). - Wire `Series.rolling()` and `DataFrame.rolling()` instance methods. - Export `Rolling`, `RollingSeriesLike`, `RollingOptions` from `src/index.ts`; export `DataFrameRolling` from `src/core/index.ts` and `src/index.ts`. - Add 40+ unit + property-based tests in `tests/window/rolling.test.ts`. - Add interactive playground page `playground/rolling.html` (8 examples). - Mark rolling as complete in `playground/index.html`. Run: https://github.com/githubnext/tsessebe/actions/runs/24004857590 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- src/window/expanding.ts: Expanding class with pandas-compatible growing-window API (mean/sum/std/var/min/max/count/median/apply). ExpandingSeriesLike interface avoids circular imports. DataFrameExpanding in frame.ts. - src/core/cat_accessor.ts: CategoricalAccessor (Series.cat) — categories, codes, ordered, nCategories, addCategories, removeCategories, removeUnusedCategories, renameCategories, setCategories, reorderCategories, asOrdered, asUnordered, valueCounts. CatHolder preserves category metadata through chaining. - Tests: 40+ unit + property tests for Expanding, 40+ unit + property tests for CategoricalAccessor. - Playground: playground/cat_accessor.html (8 tutorial sections). - Metric: 16 → 18 (+2) Run: https://github.com/githubnext/tsessebe/actions/runs/24005927691 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- src/reshape/melt.ts: pandas-compatible melt() (wide→long unpivot) with id_vars, value_vars, var_name, value_name, ignore_index - src/reshape/pivot.ts: pandas-compatible pivot() (unique reshape) and pivotTable() (aggregation — mean/sum/count/min/max/first/last) with fill_value, dropna - tests/reshape/melt.test.ts: 12 unit tests + 2 property tests (fast-check) - tests/reshape/pivot.test.ts: 14 unit tests + 2 property tests (fast-check) - playground/melt.html, playground/pivot.html: interactive tutorials - Metric: 18 → 20 (pandas_features_ported) Run: https://github.com/githubnext/tsessebe/actions/runs/24006370785 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Implements src/window/ewm.ts with: - EWM class: mean (adjust=true/false), std, var, cov, corr, apply - Decay via span, com, halflife, or alpha - ignoreNa option (false=decay-on-NaN, true=skip-NaN) - minPeriods support - DataFrameEwm in core/frame.ts for per-column EWM on DataFrames - Series.ewm() and DataFrame.ewm() methods Online weighted-sum formulation (S, W, W2 state) for O(n) computation. Helper functions computeCov/computeCorr reduce cognitive complexity. 55+ tests in tests/window/ewm.test.ts (unit + property-based). Playground: playground/ewm.html (9 sections). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Implements stack/unstack as complementary inverse operations:
- stack(df): rotates column labels into compound string index labels
("rowLabel|colName"), returning a Series<Scalar>. Supports dropna
(default true, matching pandas) and custom separator.
- unstack(series): reverses stack by parsing compound labels, filling
missing (row, col) combinations with fill_value (default null).
Both functions are standalone (no circular dep), following the melt/pivot
pattern. Round-trips correctly: unstack(stack(df, {dropna:false})) ≈ df.
Files added:
- src/reshape/stack_unstack.ts — implementation
- tests/reshape/stack_unstack.test.ts — 25+ unit + 4 property tests
- playground/stack_unstack.html — 6-section interactive tutorial
Metric: 21 → 22 (+1)
Run: https://github.com/githubnext/tsessebe/actions/runs/24007420044
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Implement pandas.MultiIndex — hierarchical multi-level index: - src/core/multi_index.ts: MultiIndex class with fromTuples/fromArrays/fromProduct - Properties: nlevels, size, names, levels, codes, isUnique, hasDuplicates - Access: at(), getLoc(), contains(), isin() - Level ops: droplevel(), swaplevel(), reorderLevels(), setNames() - Missing values: isna(), notna(), dropna() - Duplicates: duplicated(), dropDuplicates() - Set ops: union(), intersection(), difference() - sortValues(), equals(), Symbol.iterator, toString() - 60+ unit tests + 7 property-based tests (fast-check) - Playground: playground/multi_index.html (8 sections) Run: https://github.com/githubnext/tsessebe/actions/runs/24008035023 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add src/stats/rank.ts implementing pandas.Series.rank() / DataFrame.rank(): - rankSeries(series, options): assign numerical ranks to Series values - rankDataFrame(df, options): rank columns (axis=0) or rows (axis=1) - Tie methods: average (default), min, max, first, dense - NaN handling: keep (→ NaN), top (lowest ranks), bottom (highest ranks) - Percentage rank: pct=true divides by nValid (keep) or n (top/bottom) Tests: 40+ unit tests + 6 property-based tests (fast-check). Playground: playground/rank.html (8 interactive sections). Metric: 23 → 24 pandas_features_ported. Run: https://github.com/githubnext/tsessebe/actions/runs/24008535770 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ection - src/stats/nlargest.ts: nlargestSeries, nsmallestSeries, nlargestDataFrame, nsmallestDataFrame - Mirror pandas Series.nlargest/nsmallest and DataFrame.nlargest/nsmallest - keep='first'/'last'/'all' tie-handling at selection boundary - NaN/null values excluded from result (same as pandas) - Multi-column DataFrame sorting with lexicographic comparison - 45+ unit tests + 5 property-based tests - Playground: nlargest.html (8 sections) Run: https://github.com/githubnext/tsessebe/actions/runs/24009034419 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Added src/stats/cum_ops.ts — cumsum(), cumprod(), cummax(), cummin() for Series - Added dataFrameCumsum(), dataFrameCumprod(), dataFrameCummax(), dataFrameCummin() - skipna option (default true): NaN/null positions return NaN/null, accumulator skips them - skipna: false: any missing value poisons all subsequent results - axis=0 (default): column-wise; axis=1: row-wise across columns - Non-numeric values treated as missing in cumsum/cumprod - cummax/cummin work on any comparable scalar (numbers, strings, booleans) - 50+ unit tests + 6 property-based tests (fast-check) - Playground: playground/cum_ops.html (8 interactive sections) Run: https://github.com/githubnext/tsessebe/actions/runs/24009460051 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…() for Series and DataFrame Run: https://github.com/githubnext/tsessebe/actions/runs/24010099827 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…taFrame - `valueCounts(series, opts)`: count or normalize unique values, with sort/ascending/dropna - `dataFrameValueCounts(df, opts)`: count unique row combinations across all or subset of columns - 25 unit + property-based tests (fast-check); all pass - Interactive playground page: playground/value_counts.html - Wire up elem_ops exports (clip/seriesAbs/seriesRound) to stats barrel and main index Run: https://github.com/githubnext/tsessebe/actions/runs/24010521196 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
🤖 This PR is maintained by Autoloop. Each accepted iteration adds a commit to this branch.
Program Goal
Build
tsb, a complete TypeScript port of pandas, one feature at a time.Metric:
pandas_features_ported(higher is better)Current best: 28 (iteration 72)
Branch:
autoloop/build-tsb-pandas-typescript-migrationIteration 72 Summary
src/stats/value_counts.ts— value_counts for Series and DataFrameMirrors
pandas.Series.value_counts()andpandas.DataFrame.value_counts():valueCounts(series, opts)— count (or proportion) of each unique value in a SeriesdataFrameValueCounts(df, opts)— count unique row combinations across selected columnsnormalize,sort,ascending,dropna,subset(DataFrame only)scalarKey()mapper produces stableMapstring keys for anyScalartype"v1|v2|…"labels for DataFrame row combinationsAlso: wire up
elem_opsbarrel exports (iter 71 gap)src/stats/index.tsandsrc/index.tsnow exportclip,seriesAbs,seriesRound,dataFrameClip,dataFrameAbs,dataFrameRound(these were added in iter 71 but not wired to the barrels).Metric: 27 → 28 (+1)
References: Run §24010521196