Top Python use cases: Bringing machine learning to the enterprise
What are developers and data scientists using Python for? The language continues to grow in the field of machine learning, web development, and neural networks. ActiveState recently examined the data from their customers and listed the top use cases by industry and what ML tools professionals are using.
ActiveState keeps a steady hand on the pulse of open source languages, releasing regular reports examining their inner workings. Python recently took the spotlight when ActiveState examined the top ten use cases for the language.
Python is growing in usage (Stack Overflow named it the “fastest-growing major programming language” and the TIOBE index crowned it language of the year in 2018). See what industries are using Python for.
Top Python use cases
According to the data sheet released by ActiveState, these are the top ten Python use cases by industry:
- Insurance: creating business insights with machine learning
- Retail banking: flexibile data transformation and manipulation
- Aerospace: meeting software system deadlines
- Finance: data mining to identity cross-sell opportunities
- Business services: API access to financial information
- Hardware: automating network administration
- Healthcare: predicting disease prognosis
- Consulting services: bespoke web development
- InfoTech: IT modernization
- Software: adding extensibility to legacy applications
This data comes from ActiveState’s customers’ usage.
Python was originally seen as “a scripting solution for sysadmins” and since expanded to more varied use cases, including machine learning, and web development.
The report notes that machine learning usage with Python is growing.
For example, healthcare professionals use Python-based neural networks to help predict patient health outcomes. Insurance companies use a data science-focused Python build that also helps protects customers from lawsuits. Customer cross-selling services receive and correlate their transaction data from machine learning.
Best ML language?
A white paper from ActiveState, “Unlocking the power of data science & machine learning with Python” explores the potentials Python holds in these fields. It is common knowledge that Python is popular within fields of scientific study.
When compared against other languages used in data science and machine learning Python offers up strong general usage stats.
For instance, its learning curve is not as steep as R, Java, or Go. Its speed varies, depending on library usage, and the language benefits from the mass amounts of deployment tools available.
However, the report calls Go a recent “alternative to Python and R as a solution to issues to issues around deployment and maintenance of data science code in production”. Go’s data science ecosystem currently lacks the breadth and depth of Python. This is most likely due to the language’s relative youth.
Recommended machine learning tools
The report goes into detail of some tools to use in conjunction with Python in machine learning use cases.
SEE ALSO: Falling into the automation honey trap
- Complementing Python with R: ActiveState suggests that data scientists “use R for its statistical functions and then wrap their model in a Python application that has a variety of additional features”.
- Data frames with Pandas: If you do not want to use R, bring over one of the features from R (data frames) with Pandas.
- Jupyter Notebooks: The browser-based tool is a commonly used medium in data science for analysis presentation and sharing.
- NLTK: The Natural Language Toolkit library provides repositories for natural language processing.
- TensorFlow: Google’s deep learning model library can power image classification, NLP services, and more.
- Keras: Keras can be used on top of TensorFlow for more control and advanced machine learning.