pandas-pro

Name: pandas-pro
Author: Jeffallan

4.2kinstalls971forks

How to install pandas-pro?

jeffallan/claude-skills/pandas-pro

$npx -y skills add Jeffallan/claude-skills --skill pandas-pro

Installs into the current project.

Summary

Performs pandas DataFrame operations for data analysis, manipulation, and transformation. Use when working with pandas DataFrames, data cleaning, aggregation, merging, or time series analysis. Invoke for data manipulation tasks such as joining DataFrames on multiple keys, pivotin

Files · 1

View on GitHub

SKILL.md

1	# Pandas Pro
2
3	Expert pandas developer specializing in efficient data manipulation, analysis, and transformation workflows with production-grade performance patterns.
4
5	## Core Workflow
6
7	1. Assess data structure — Examine dtypes, memory usage, missing values, data quality:
8	```python
9	print(df.dtypes)
10	print(df.memory_usage(deep=True).sum() / 1e6, "MB")
11	print(df.isna().sum())
12	print(df.describe(include="all"))
13	```
14	2. Design transformation — Plan vectorized operations, avoid loops, identify indexing strategy
15	3. Implement efficiently — Use vectorized methods, method chaining, proper indexing
16	4. Validate results — Check dtypes, shapes, null counts, and row counts:
17	```python
18	assert result.shape[0] == expected_rows, f"Row count mismatch: {result.shape[0]}"
19	assert result.isna().sum().sum() == 0, "Unexpected nulls after transform"
20	assert set(result.columns) == expected_cols
21	```
22	5. Optimize — Profile memory, apply categorical types, use chunking if needed
23
24	## Reference Guide
25
26	Load detailed guidance based on context:
27
28	\| Topic \| Reference \| Load When \|
29	\|-------\|-----------\|-----------\|
30	\| DataFrame Operations \| `references/dataframe-operations.md` \| Indexing, selection, filtering, sorting \|
31	\| Data Cleaning \| `references/data-cleaning.md` \| Missing values, duplicates, type conversion \|
32	\| Aggregation & GroupBy \| `references/aggregation-groupby.md` \| GroupBy, pivot, crosstab, aggregation \|
33	\| Merging & Joining \| `references/merging-joining.md` \| Merge, join, concat, combine strategies \|
34	\| Performance Optimization \| `references/performance-optimization.md` \| Memory usage, vectorization, chunking \|
35
36	## Code Patterns
37
38	### Vectorized Operations (before/after)
39
40	```python
41	# ❌ AVOID: row-by-row iteration
42	for i, row in df.iterrows():
43	df.at[i, 'tax'] = row['price'] * 0.2
44
45	# ✅ USE: vectorized assignment
46	df['tax'] = df['price'] * 0.2
47	```
48
49	### Safe Subsetting with `.copy()`
50
51	```python
52	# ❌ AVOID: chained indexing triggers SettingWithCopyWarning
53	df['A']['B'] = 1
54
55	# ✅ USE: .loc[] with explicit copy when mutating a subset
56	subset = df.loc[df['status'] == 'active', :].copy()
57	subset['score'] = subset['score'].fillna(0)
58	```
59
60	### GroupBy Aggregation
61
62	```python
63	summary = (
64	df.groupby(['region', 'category'], observed=True)
65	.agg(
66	total_sales=('revenue', 'sum'),
67	avg_price=('price', 'mean'),
68	order_count=('order_id', 'nunique'),
69	)
70	.reset_index()
71	)
72	```
73
74	### Merge with Validation
75
76	```python
77	merged = pd.merge(
78	left_df, right_df,
79	on=['customer_id', 'date'],
80	how='left',
81	validate='m:1', # asserts right key is unique
82	indicator=True,
83	)
84	unmatched = merged[merged['_merge'] != 'both']
85	print(f"Unmatched rows: {len(unmatched)}")
86	merged.drop(columns=['_merge'], inplace=True)
87	```
88
89	### Missing Value Handling
90
91	```python
92	# Forward-fill then interpolate numeric gaps
93	df['price'] = df['price'].ffill().interpolate(method='linear')
94
95	# Fill categoricals with mode, numerics with median
96	for col in df.select_dtypes(include='object'):
97	df[col] = df[col].fillna(df[col].mode()[0])
98	for col in df.select_dtypes(include='number'):
99	df[col] = df[col].fillna(df[col].median())
100	```
101
102	### Time Series Resampling
103
104	```python
105	daily = (
106	df.set_index('timestamp')
107	.resample('D')
108	.agg({'revenue': 'sum', 'sessions': 'count'})
109	.fillna(0)
110	)
111	```
112
113	### Pivot Table
114
115	```python
116	pivot = df.pivot_table(
117	values='revenue',
118	index='region',
119	columns='product_line',
120	aggfunc='sum',
121	fill_value=0,
122	margins=True,
123	)
124	```
125
126	### Memory Optimization
127
128	```python
129	# Downcast numerics and convert low-cardinality strings to categorical
130	df['category'] = df['category'].astype('category')
131	df['count'] = pd.to_numeric(df['count'], downcast='integer')
132	df['score'] = pd.to_numeric(df['score'], downcast='float')
133	print(df.memory_usage(deep=True).sum() / 1e6, "MB after optimization")
134	```
135
136	## Constraints
137
138	### MUST DO
139	- Use vectorized operations instead of loops
140	- Set appropriate dtypes (categorical for low-cardinality strings)
141	- Check memory usage with `.memory_usage(deep=True)`
142	- Handle missing values explicitly (don't silently drop)
143	- Use method

People who liked this also viewed

6 picks

Type

pandas-pro

4.2kinstalls971forks

View on GitHub

How to install pandas-pro?

jeffallan/claude-skills/pandas-pro

$npx -y skills add Jeffallan/claude-skills --skill pandas-pro

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Pandas Pro
2
3	Expert pandas developer specializing in efficient data manipulation, analysis, and transformation workflows with production-grade performance patterns.
4
5	## Core Workflow
6
7	1. Assess data structure — Examine dtypes, memory usage, missing values, data quality:
8	```python
9	print(df.dtypes)
10	print(df.memory_usage(deep=True).sum() / 1e6, "MB")
11	print(df.isna().sum())
12	print(df.describe(include="all"))
13	```
14	2. Design transformation — Plan vectorized operations, avoid loops, identify indexing strategy
15	3. Implement efficiently — Use vectorized methods, method chaining, proper indexing
16	4. Validate results — Check dtypes, shapes, null counts, and row counts:
17	```python
18	assert result.shape[0] == expected_rows, f"Row count mismatch: {result.shape[0]}"
19	assert result.isna().sum().sum() == 0, "Unexpected nulls after transform"
20	assert set(result.columns) == expected_cols
21	```
22	5. Optimize — Profile memory, apply categorical types, use chunking if needed
23
24	## Reference Guide
25
26	Load detailed guidance based on context:
27
28	\| Topic \| Reference \| Load When \|
29	\|-------\|-----------\|-----------\|
30	\| DataFrame Operations \| `references/dataframe-operations.md` \| Indexing, selection, filtering, sorting \|
31	\| Data Cleaning \| `references/data-cleaning.md` \| Missing values, duplicates, type conversion \|
32	\| Aggregation & GroupBy \| `references/aggregation-groupby.md` \| GroupBy, pivot, crosstab, aggregation \|
33	\| Merging & Joining \| `references/merging-joining.md` \| Merge, join, concat, combine strategies \|
34	\| Performance Optimization \| `references/performance-optimization.md` \| Memory usage, vectorization, chunking \|
35
36	## Code Patterns
37
38	### Vectorized Operations (before/after)
39
40	```python
41	# ❌ AVOID: row-by-row iteration
42	for i, row in df.iterrows():
43	df.at[i, 'tax'] = row['price'] * 0.2
44
45	# ✅ USE: vectorized assignment
46	df['tax'] = df['price'] * 0.2
47	```
48
49	### Safe Subsetting with `.copy()`
50
51	```python
52	# ❌ AVOID: chained indexing triggers SettingWithCopyWarning
53	df['A']['B'] = 1
54
55	# ✅ USE: .loc[] with explicit copy when mutating a subset
56	subset = df.loc[df['status'] == 'active', :].copy()
57	subset['score'] = subset['score'].fillna(0)
58	```
59
60	### GroupBy Aggregation
61
62	```python
63	summary = (
64	df.groupby(['region', 'category'], observed=True)
65	.agg(
66	total_sales=('revenue', 'sum'),
67	avg_price=('price', 'mean'),
68	order_count=('order_id', 'nunique'),
69	)
70	.reset_index()
71	)
72	```
73
74	### Merge with Validation
75
76	```python
77	merged = pd.merge(
78	left_df, right_df,
79	on=['customer_id', 'date'],
80	how='left',
81	validate='m:1', # asserts right key is unique
82	indicator=True,
83	)
84	unmatched = merged[merged['_merge'] != 'both']
85	print(f"Unmatched rows: {len(unmatched)}")
86	merged.drop(columns=['_merge'], inplace=True)
87	```
88
89	### Missing Value Handling
90
91	```python
92	# Forward-fill then interpolate numeric gaps
93	df['price'] = df['price'].ffill().interpolate(method='linear')
94
95	# Fill categoricals with mode, numerics with median
96	for col in df.select_dtypes(include='object'):
97	df[col] = df[col].fillna(df[col].mode()[0])
98	for col in df.select_dtypes(include='number'):
99	df[col] = df[col].fillna(df[col].median())
100	```
101
102	### Time Series Resampling
103
104	```python
105	daily = (
106	df.set_index('timestamp')
107	.resample('D')
108	.agg({'revenue': 'sum', 'sessions': 'count'})
109	.fillna(0)
110	)
111	```
112
113	### Pivot Table
114
115	```python
116	pivot = df.pivot_table(
117	values='revenue',
118	index='region',
119	columns='product_line',
120	aggfunc='sum',
121	fill_value=0,
122	margins=True,
123	)
124	```
125
126	### Memory Optimization
127
128	```python
129	# Downcast numerics and convert low-cardinality strings to categorical
130	df['category'] = df['category'].astype('category')
131	df['count'] = pd.to_numeric(df['count'], downcast='integer')
132	df['score'] = pd.to_numeric(df['score'], downcast='float')
133	print(df.memory_usage(deep=True).sum() / 1e6, "MB after optimization")
134	```
135
136	## Constraints
137
138	### MUST DO
139	- Use vectorized operations instead of loops
140	- Set appropriate dtypes (categorical for low-cardinality strings)
141	- Check memory usage with `.memory_usage(deep=True)`
142	- Handle missing values explicitly (don't silently drop)
143	- Use method

People who liked this also viewed

6 picks

Type