distogram

3.0.3 verified Fri May 01 auth: no python

A library to compute histograms on distributed environments, on streaming data. Current version: 3.0.3. Released occasionally, with breaking changes from v2 to v3 (API redesign).

pip install distogram

Common errors

error AttributeError: module 'distogram' has no attribute 'distogram' ↓

cause Trying to use the old v2 API (distogram.distogram) on v3.

fix

Use from distogram import Distogram and then Distogram().

error TypeError: distogram() takes 1 positional argument but 2 were given ↓

cause Calling Distogram(bin_count=50) which is v2 syntax. v3 uses max_bins.

fix

Use Distogram(max_bins=50).

error ImportError: cannot import name 'Distogram' from 'distogram' ↓

cause Installed distogram v2.x which does not have the class. Or typo.

fix

Upgrade to distogram>=3.0.0 with pip install --upgrade distogram.

Warnings

breaking distogram v3 changes the API from a function-based interface to a class-based one. The old distogram() function is replaced by the Distogram class. Functions like insert, histogram, merge are now module-level functions taking a Distogram object as first argument. ↓

fix Replace distogram() with Distogram(), and use module-level functions: from distogram import Distogram, insert, histogram.

breaking distogram v3 removes the bin_count parameter from Distogram constructor. Use max_bins instead. ↓

fix Change Distogram(bin_count=50) to Distogram(max_bins=50).

deprecated The old distogram module function (lowercase) is removed in v3. Trying to import it will raise ImportError. ↓

fix Use from distogram import Distogram (capital D).

gotcha distogram does not preserve precise quantiles; it's an approximation. Do not rely on exact percentile values. ↓

fix Use for approximate histograms only; for exact quantiles use other tools.

Imports

Distogram
wrong
```
from distogram import distogram
```
correct
```
from distogram import Distogram
```
In distogram v3, the class is named Distogram (capital D). v2 used a lowercase distogram function.
insert
wrong
```
distogram.insert(dist, value)
```
correct
```
from distogram import insert
```
The module-level insert function is now importable directly.
histogram
wrong
```
distogram.histogram(dist, bins=10)
```
correct
```
from distogram import histogram
```
The histogram function is now separate; in v2 it was method on distogram object.

Quickstart

Create a Distogram, insert data points, then retrieve histogram bins.

from distogram import Distogram, insert, histogram

dist = Distogram()
for v in [1, 2, 2, 3, 3, 3, 4, 4, 4, 4]:
    insert(dist, v)
bins = histogram(dist, bins=3)
print(bins)