Iteration 48: Add 29 new benchmark pairs (rolling variants, expanding, ewm, series ops, categorical, DataFrame ops)

Run: https://github.com/githubnext/tsessebe/actions/runs/24368561859

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: github-actions[bot]

benchmarks/pandas/bench_cat_from_codes.py

@@ -0,0 +1,21 @@
+"""Benchmark: Categorical from codes on 100k-element array"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+categories = ["apple", "banana", "cherry", "date", "elderberry"]
+codes = np.arange(ROWS) % len(categories)
+
+for _ in range(WARMUP):
+    pd.Categorical.from_codes(codes, categories=categories)
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    pd.Categorical.from_codes(codes, categories=categories)
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "cat_from_codes", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_cut.py

@@ -0,0 +1,21 @@
+"""Benchmark: cut (bin into 10 bins) on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = (np.arange(ROWS) % 10000) * 0.01
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    pd.cut(s, 10)
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    pd.cut(s, 10)
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "cut", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_dataframe_cov.py

@@ -0,0 +1,22 @@
+"""Benchmark: DataFrame covariance matrix on 1000x10 DataFrame"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 1_000
+COLS = 10
+WARMUP = 3
+ITERATIONS = 10
+
+data = {f"col{c}": np.sin(np.arange(ROWS) * 0.01 + c) for c in range(COLS)}
+df = pd.DataFrame(data)
+
+for _ in range(WARMUP):
+    df.cov()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    df.cov()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "dataframe_cov", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_ewm_std.py

@@ -0,0 +1,21 @@
+"""Benchmark: ewm std (alpha=0.1) on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = np.sin(np.arange(ROWS) * 0.01)
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    s.ewm(alpha=0.1).std()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.ewm(alpha=0.1).std()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "ewm_std", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_ewm_var.py

@@ -0,0 +1,21 @@
+"""Benchmark: ewm var (alpha=0.1) on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = np.sin(np.arange(ROWS) * 0.01)
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    s.ewm(alpha=0.1).var()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.ewm(alpha=0.1).var()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "ewm_var", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_expanding_std.py

@@ -0,0 +1,21 @@
+"""Benchmark: expanding std on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = np.sin(np.arange(ROWS) * 0.01)
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    s.expanding().std()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.expanding().std()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "expanding_std", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_expanding_sum.py

@@ -0,0 +1,21 @@
+"""Benchmark: expanding sum on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = np.sin(np.arange(ROWS) * 0.01)
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    s.expanding().sum()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.expanding().sum()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "expanding_sum", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_expanding_var.py

@@ -0,0 +1,21 @@
+"""Benchmark: expanding var on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = np.sin(np.arange(ROWS) * 0.01)
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    s.expanding().var()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    s.expanding().var()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "expanding_var", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_insert_column.py

@@ -0,0 +1,29 @@
+"""Benchmark: DataFrame insert column on 10000x3 DataFrame"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 10_000
+WARMUP = 3
+ITERATIONS = 20
+
+new_col = np.arange(ROWS, dtype=float) * 4
+
+def make_df():
+    return pd.DataFrame({
+        "a": np.arange(ROWS, dtype=float),
+        "b": np.arange(ROWS, dtype=float) * 2,
+        "c": np.arange(ROWS, dtype=float) * 3,
+    })
+
+for _ in range(WARMUP):
+    df = make_df()
+    df.insert(1, "new_col", new_col)
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    df = make_df()
+    df.insert(1, "new_col", new_col)
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "insert_column", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))

benchmarks/pandas/bench_qcut.py

@@ -0,0 +1,21 @@
+"""Benchmark: qcut (10 quantile bins) on 100k-element Series"""
+import json, time
+import numpy as np
+import pandas as pd
+
+ROWS = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+data = (np.arange(ROWS) % 10000) * 0.01
+s = pd.Series(data)
+
+for _ in range(WARMUP):
+    pd.qcut(s, 10, duplicates="drop")
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    pd.qcut(s, 10, duplicates="drop")
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({ "function": "qcut", "mean_ms": total / ITERATIONS, "iterations": ITERATIONS, "total_ms": total }))