prep for TR

Author: iansedano

pandas-iterate-over-rows/README.md

@@ -1,3 +1,11 @@
-# Placeholder
+# How to Iterate Over a pandas DataFrame's Rows
 
-Info...
+In these supplementary materials you'll find the examples discussed in the tutorial, along with some bonus materials.
+
+The main extra dependency for the bonus materials is [perfplot](https://github.com/nschloe/perfplot). To install all dependencies into your venv
+
+```shell
+$ python -m pip install pandas httpx codetiming perfplot
+```
+
+You'll also find some additional code samples along the way, showing alternatives to looping, such as by using list comprehensions, and even alternative ways of looping.

…iterate-over-rows/time_cumulative_sum.py …e-over-rows/cumulative_sum_codetiming.pypandas-iterate-over-rows/time_cumulative_sum.py renamed to pandas-iterate-over-rows/cumulative_sum_codetiming.py pandas-iterate-over-rows/time_cumulative_sum.py renamed to pandas-iterate-over-rows/cumulative_sum_codetiming.py

@@ -2,31 +2,28 @@
 import pandas as pd
 
 
-def pandas_cumsum(products):
-    return products.assign(
-        income=lambda df: df["sales"] * df["unit_price"],
-        cumulative_income=lambda df: df["income"].cumsum(),
-    ).drop(columns="income")
-
-
 def loop_cumsum(products):
     cumulative_sum = []
-
     for row in products.itertuples():
         if cumulative_sum:
             cumulative_sum.append(
                 cumulative_sum[-1] + (row.sales * row.unit_price)
             )
         else:
             cumulative_sum.append(row.sales * row.unit_price)
-
     return products.assign(cumulative_income=cumulative_sum)
 
 
+def pandas_cumsum(products):
+    return products.assign(
+        income=lambda df: df["sales"] * df["unit_price"],
+        cumulative_income=lambda df: df["income"].cumsum(),
+    ).drop(columns="income")
+
+
 for f in [pandas_cumsum, loop_cumsum]:
     products = pd.read_csv("resources/products.csv")
-    products = pd.concat(products for _ in range(200))
-
+    products = pd.concat(products for _ in range(1000))
     with codetiming.Timer(
         name=f.__name__, text="{name:20}: {milliseconds:.2f} ms"
     ):

pandas-iterate-over-rows/cumulative_sum_perfplot.py

@@ -0,0 +1,35 @@
+import pandas as pd
+import perfplot
+
+
+def loop_cumsum(products):
+    cumulative_sum = []
+    for row in products.itertuples():
+        if cumulative_sum:
+            cumulative_sum.append(
+                cumulative_sum[-1] + (row.sales * row.unit_price)
+            )
+        else:
+            cumulative_sum.append(row.sales * row.unit_price)
+    return products.assign(cumulative_income=cumulative_sum)
+
+
+def pandas_cumsum(products):
+    return products.assign(
+        income=lambda df: df["sales"] * df["unit_price"],
+        cumulative_income=lambda df: df["income"].cumsum(),
+    ).drop(columns="income")
+
+
+products = pd.read_csv("resources/products.csv")
+
+plot = perfplot.bench(
+    n_range=[i**2 for i in range(1, 1000, 100)],
+    setup=lambda n: pd.concat([products for _ in range(n)]),
+    kernels=[pandas_cumsum, loop_cumsum],
+    labels=["pandas cumsum", "loop cumsum"],
+    equality_check=None,
+)
+
+plot.show()
+plot.show(logy=True)

pandas-iterate-over-rows/examples.py

@@ -67,7 +67,7 @@ def get_books(n):
     return pd.concat([books for _ in range((n // len(books)) + 1)]).iloc[:n]
 
 
-perfplot.live(
+plot = perfplot.bench(
     setup=lambda n: get_books(n),
     kernels=[
         clean_pub_replace,
@@ -79,5 +79,7 @@ def get_books(n):
     labels=["replace", "itertuples", "iterrows", "apply", "list comp"],
     n_range=[i**2 for i in range(1, 40, 2)],
     equality_check=None,
-    logy=True,
 )
+
+plot.show()
+plot.show(logy=True)

…as-iterate-over-rows/fix_place_of_pub.py …e-over-rows/fix_place_of_pub_perfplot.pypandas-iterate-over-rows/fix_place_of_pub.py renamed to pandas-iterate-over-rows/fix_place_of_pub_perfplot.py pandas-iterate-over-rows/fix_place_of_pub.py renamed to pandas-iterate-over-rows/fix_place_of_pub_perfplot.py

@@ -19,6 +19,7 @@ def python_sum(webs):
 
 for f in [pandas_sum, loop_sum, python_sum]:
     webs = pd.read_csv("resources/popular_websites.csv", index_col=0)
+    webs = pd.concat([webs for _ in range(1000)])
     with codetiming.Timer(
         name=f.__name__, text="{name:20}: {milliseconds:.2f} ms"
     ):

pandas-iterate-over-rows/generate_data.py

@@ -19,11 +19,12 @@ def python_sum(webs):
 
 webs = pd.read_csv("resources/popular_websites.csv", index_col=0)
 
-perfplot.live(
+plot = perfplot.bench(
     n_range=[i**2 for i in range(1, 1000, 100)],
     setup=lambda n: pd.concat([webs for _ in range(n)]),
     kernels=[pandas_sum, loop_sum, python_sum],
     labels=["pandas sum", "loop sum", "python sum"],
-    # equality_check=None,
-    logy=True,
 )
+
+plot.show()
+plot.show(logy=True)