import pandas as pd
import polars as pl
from mis_analytics.core import clean_col_names
from mis_analytics.etl import trackmis_analytics
mis_analytics helps prepare supply chain and manufacturing data. Clean columns and resample time series, inspect pandas and eager Polars pipelines, and use deSAPher to look up SAP fields and convert SAP extracts.
Installation
Requires Python 3.11 or later. Install the DataFrame utilities and pipeline tracking:
pip install mis_analyticsFor SAP functionality, install the desapher extra:
pip install "mis_analytics[desapher]"The extra adds httpx, beautifulsoup4, and lxml. The base package includes pandas, NumPy, and Polars. Import mis_analytics.desapher only when its extra dependencies are installed.
To install the latest code from GitHub, including the extra:
pip install "mis_analytics[desapher] @ git+https://github.com/MIS-Analytics/mis_analytics.git"Omit [desapher] for the base package.
DataFrame utilities
orders = pd.DataFrame({'Order ID': [101, 102, 103], 'Quantity': [5, 10, 15]})
orders = clean_col_names(orders)
orders| order_id | quantity | |
|---|---|---|
| 0 | 101 | 5 |
| 1 | 102 | 10 |
| 2 | 103 | 15 |
Pipeline tracking
Decorate a function with @track and chain it with .pipe(). Set vrbs=True on each step to print its name, docstring, elapsed time, shape changes, and input/output samples. Logging is off by default. Both pandas and eager Polars DataFrames are supported; each transformation must support the DataFrame it receives.
@track
def take_rows(df, n):
"""Keep the first n rows."""
return df.head(n)pandas_preview = orders.pipe(take_rows, 2, vrbs=True).pipe(take_rows, 1, vrbs=True)*************** take_rows ***************
'''Keep the first n rows.'''
Total Time: 691 µs
Start: 2026-09-16 11:42:40.985041
End: 2026-09-16 11:42:40.985732
Input DataFrame:
head:
order_id quantity
0 101 5
1 102 10
2 103 15
sample:
order_id quantity
1 102 10
2 103 15
0 101 5
tail:
order_id quantity
0 101 5
1 102 10
2 103 15
Input: 3 rows, 2 cols
↓ ↓
Diff: -1 rows, 0 cols
↓ ↓
Output: 2 rows, 2 cols
Output DataFrame:
head:
order_id quantity
0 101 5
1 102 10
sample:
order_id quantity
0 101 5
1 102 10
tail:
order_id quantity
0 101 5
1 102 10
*************** take_rows ***************
'''Keep the first n rows.'''
Total Time: 217 µs
Start: 2026-09-16 11:42:40.989443
End: 2026-09-16 11:42:40.989660
Input DataFrame:
head:
order_id quantity
0 101 5
1 102 10
sample:
order_id quantity
1 102 10
0 101 5
tail:
order_id quantity
0 101 5
1 102 10
Input: 2 rows, 2 cols
↓ ↓
Diff: -1 rows, 0 cols
↓ ↓
Output: 1 rows, 2 cols
Output DataFrame:
head:
order_id quantity
0 101 5
sample:
order_id quantity
0 101 5
tail:
order_id quantity
0 101 5
polars_orders = pl.DataFrame({'order_id': [101, 102, 103], 'quantity': [5, 10, 15]})
polars_preview = polars_orders.pipe(take_rows, 2, vrbs=True).pipe(take_rows, 1, vrbs=True)*************** take_rows ***************
'''Keep the first n rows.'''
Total Time: 257 µs
Start: 2026-09-16 11:42:58.817200
End: 2026-09-16 11:42:58.817457
Input DataFrame:
head:
shape: (3, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
│ 103 ┆ 15 │
└──────────┴──────────┘
sample:
shape: (3, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
│ 103 ┆ 15 │
└──────────┴──────────┘
tail:
shape: (3, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
│ 103 ┆ 15 │
└──────────┴──────────┘
Input: 3 rows, 2 cols
↓ ↓
Diff: -1 rows, 0 cols
↓ ↓
Output: 2 rows, 2 cols
Output DataFrame:
head:
shape: (2, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
└──────────┴──────────┘
sample:
shape: (2, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
└──────────┴──────────┘
tail:
shape: (2, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
└──────────┴──────────┘
*************** take_rows ***************
'''Keep the first n rows.'''
Total Time: 328 µs
Start: 2026-09-16 11:42:58.817901
End: 2026-09-16 11:42:58.818229
Input DataFrame:
head:
shape: (2, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
└──────────┴──────────┘
sample:
shape: (2, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
└──────────┴──────────┘
tail:
shape: (2, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
│ 102 ┆ 10 │
└──────────┴──────────┘
Input: 2 rows, 2 cols
↓ ↓
Diff: -1 rows, 0 cols
↓ ↓
Output: 1 rows, 2 cols
Output DataFrame:
head:
shape: (1, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
└──────────┴──────────┘
sample:
shape: (1, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
└──────────┴──────────┘
tail:
shape: (1, 2)
┌──────────┬──────────┐
│ order_id ┆ quantity │
│ --- ┆ --- │
│ i64 ┆ i64 │
╞══════════╪══════════╡
│ 101 ┆ 5 │
└──────────┴──────────┘
SAP helpers with deSAPher
With the desapher extra installed, fetch SAP table metadata from sapdatasheet.org and use it to convert types and rename fields in pandas extracts. Fetching metadata requires internet access.
from mis_analytics.desapher import get_sap_tables_structure, convert_sap_types, rename_sap_columns
sap_sheet = get_sap_tables_structure(['MARA'])
materials = pd.DataFrame({'MATNR': ['000000000000000101']})
if sap_sheet is not None:
materials = materials.pipe(convert_sap_types, sap_sheet).pipe(rename_sap_columns, sap_sheet)get_sap_table_description also retrieves a table’s description.
Development
From a local checkout, install the development group and SAP extra:
uv sync --group dev --extra desapherEdit the source notebooks under nbs/, then export the modules. Regenerate the README after editing nbs/index.ipynb:
uv run nbdev-export
uv run nbdev-test --path nbs/index.ipynb
uv run nbdev-readme