Skip to main content

natural language processing primitives for Featuretools

Project description

NLP Primitives

Tests PyPI Version Anaconda Version StackOverflow PyPI Downloads

nlp_primitives is a Python library with Natural Language Processing Primitives, intended for use with Featuretools.

nlp_primitives allows you to make use of text data in your machine learning pipeline in the same pipeline as the rest of your data.


There are two options for installing nlp_primitives. Both of the options will also install Featuretools if it is not already installed.

The first option is to install a version of nlp_primitives that does not include Tensorflow. With this option, primitives that depend on Tensorflow cannot be used. Currently, the only primitive that can not be used with this install option is UniversalSentenceEncoder.


nlp_primitives without Tensorflow can be installed with pip:

python -m pip install nlp_primitives


or from the conda-forge channel on conda:

conda install -c conda-forge nlp-primitives

The second option is to install the complete version of nlp_primitives, which will also install Tensorflow and allow use of all primitives.

To install the complete version of nlp_primitives with pip:

python -m pip install "nlp_primitives[complete]"

or from the conda-forge channel on conda:

conda install -c conda-forge nlp-primitives-complete


Calculating Features

With nlp_primitives primtives in featuretools, this is how to calculate the same feature.

from featuretools.nlp_primitives import PolarityScore

data = ["hello, this is a new featuretools library",
        "this will add new natural language primitives",
        "we hope you like it!"]

pol = PolarityScore()
0    0.365
1    0.385
2    1.000
dtype: float64

Combining Primitives

In featuretools, this is how to combine nlp_primitives primitives with built-in or other installed primitives.

import featuretools as ft
from featuretools.nlp_primitives import TitleWordCount
from featuretools.primitives import Mean

entityset = ft.demo.load_retail()
feature_matrix, features = ft.dfs(entityset=entityset, target_dataframe_name='products', agg_primitives=[Mean], trans_primitives=[TitleWordCount])

           MEAN(order_products.quantity)  MEAN(order_products.unit_price)  MEAN(  TITLE_WORD_COUNT(description)
10002                         16.795918                          1.402500                   23.556276                           3.0
10080                         13.857143                          0.679643                    8.989357                           3.0
10120                          6.620690                          0.346500                    2.294069                           2.0
10123C                         1.666667                          1.072500                    1.787500                           3.0
10124A                           3.2000                            0.6930                      2.2176                           5.0


To install from source, clone this repo and run

make installdeps-test

This will install all pip dependencies.

Built at Alteryx

NLP Primitives is an open source project maintained by Alteryx. To see the other open source projects we’re working on visit Alteryx Open Source. If building impactful data science pipelines is important to you or your business, please get in touch.

Alteryx Open Source

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

nlp_primitives-2.13.0.tar.gz (44.2 MB view hashes)

Uploaded Source

Built Distribution

nlp_primitives-2.13.0-py3-none-any.whl (44.7 MB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page