Press "Enter" to skip to content

Category: Data Science

Quantifying Model Uncertainty with Tensorflow Probability

Published 2022-04-29 by Kevin Feasel

Vini Jaiswal reviews the Tensorflow Probability library:

In this blog, we look at the topic of uncertainty quantification for machine learning and deep learning. By no means is this a new subject, but the introduction of tools such as Tensorflow Probability and Pyro have made it easy to perform probabilistic modeling to streamline uncertainty calculations. Consider the scenario in which we predict the value of an asset like a house, based on a number of features, to drive purchasing decisions. Wouldn’t it be beneficial to know how certain we are of these predicted prices? Tensorflow Probability allows you to use the familiar Tensorflow syntax and methodology but adds the ability to work with distributions. In this introductory post, we leave the priors and the Bayesian treatment behind and opt for a simpler probabilistic treatment to illustrate the basic principles. We use the likelihood principle to illustrate how an uncertainty measure can be obtained along with predicted values by applying them to a deep learning regression problem.

Read on for an interesting explanation and tutorial.

Comments closed

Custom Model Evaluation Metrics with MLflow

Published 2022-04-22 by Kevin Feasel

Mark Zhang shows off a new bit of functionality in MLflow:

According to an internal customer survey, 75% of respondents say they frequently or always use specialized, business-focused metrics in addition to basic ones like accuracy and loss. Data scientists often utilize these custom metrics as they are more descriptive of business objectives (e.g. conversion rate), and contain additional heuristics not captured by the model prediction itself.

In this blog, we introduce an easy and convenient way of evaluating MLflow models on user-defined custom metrics. With this functionality, a data scientist can easily incorporate this logic at the model evaluation stage and quickly determine the best-performing model without further downstream analysis

Click through to see how to use built-in metrics but also how to create your own.

Comments closed

Data Profiling in Python

Published 2022-04-05 by Kevin Feasel

Brendan Tierney uses Python to look at some data:

One of the most common, and sometimes boring, task when working with datasets is writing some code to profile the data. Most data scientists will have built a set of tools/scripts to help them with this regular and slightly boring task. As with most IT tasks we should be trying to automate what we can, to allow us to spend more time on more important tasks, such as deriving insights and delivering value to the business, instead of repeatedly writing code to produce various statistics about the data and drawing pretty pictures.

I’ve written previously about automating and using some data profiling libraries to help us with this task. There are lots of packages available on pypi.og and on GitHub. Below I give examples of 5 Python Data Profiling libraries, with links to their GitHubs.

Brendan includes some good examples of libraries here so check it out.

Comments closed

“Production” in Data Analytics

Published 2022-03-22 by Kevin Feasel

Joey Jablonski brings up an important point:

Data-driven environments have a fundamentally different set of needs around testing, deployment, and visibility then traditional business applications. Data driven environments need access to fresh data on a high level of update frequency to ensure that data engineers and data scientists are able to effect outputs and recommendations on a timeline that has a positive impact on business decisions and customer experiences.

My day job involves running a predictive analytics team. We train models on production data—there’s very little value in training models on artificial dev data (outside of understanding the parameters of the modeling process), so even our development data generally comes from production. I don’t know that I’m sold on data mesh as a solution to this but it’s worth investigation.

Comments closed

Snowflake Purchases Streamlit

Published 2022-03-10 by Kevin Feasel

Alex Woodie reports on a purchase:

Cloud data warehousing giant Snowflake showed it’s serious about Python and data science this week when it announced that it plans to spend $800 million to buy Streamlit, a provider of Python-based tools for rapidly developing interactive data applications on the Web.

Co-founded in San Francisco in 2018 by Adrien Treuille, Amanda Kelly, and Thiago Teixeira, Streamlit develops an open source framework of the same name that allows data scientists and machine learning engineers to create and deploy data applications. The software is compatible with other Python-based frameworks, such as NumPy, Pandas, Matplotlib, and Scikit-learn, and uses React to render screens on the front-end.

Streamlit is nice. $800 million nice? That’s a good question.

Comments closed

Types of Regression

Published 2022-02-17 by Kevin Feasel

The Finnstats folks talk about regression:

Basically, Regression analysis involves creating an equation to describe the significant association between one or more predictors and response variables, as well as estimating current observations.

The results of the regression reveal the direction, size, and analytical significance of the relationship between predictor and response, where the dependent variable is either numerical or discrete.

Click through for details on six types of regression. H/T R-Bloggers.

Comments closed

A Bayesian Approach to Pricing

Published 2022-02-15 by Kevin Feasel

John Cook has some thoughts on an interesting problem:

Suppose you want to determine how to price a product and you initially don’t know what the market is willing to pay. This post outlines some of the things you might think about, and how Bayesian modeling might help.

This post is not the final word on the subject, or even my final word on the subject. It is essentially a reply to a friend’s question turned into a blog post rather than an email.

This is a really interesting problem. Price is ultimately a signal rather than the solution. This is an attempt to understand how you start to build up enough information to start setting effective prices. Over a long enough timeframe, we’ll eventually land on a commonly accepted price but that’s an economist’s way of hand-waving away the process.

Comments closed

Finding Performance Regressions

Published 2022-01-27 by Kevin Feasel

Angus Croll embraces the power of control charts:

Technically, “performance” metrics are those relating to the responsiveness or latency of the app, including start up time.

But TV devices also tend to be more memory constrained than other devices, and as such are more liable to crash during a memory spike — so for Netflix TV we actually care about memory at least as much as performance, maybe more so.

At Netflix the term “performance” usually encompasses both performance metrics (in the strict meaning) and memory metrics, and that’s how we’re using the term here.

Click through for a primer on how they use changepoint and anomaly detection to do this.

Comments closed

Handling Categorical Data in R

Published 2022-01-13 by Kevin Feasel

The RSquared Academy blog has a two-parter on handling categorical data in R. Part 1 elaborates on kinds of categorical data and introduces a case study:

While we can rank the categories, we cannot assign a value to them. For example, in satisfaction ranking, we cannot say that like is twice as positive as dislike i.e. we are unable to say how much they differ from each other. While the order or rank of data is meaningful, the difference between two pieces of data cannot be measured/determined or are meaningless. Ordinal data provide information about relative comparisons, but not the magnitude of the differences.

Part 2 shows off ways to work with categorical data in tables:

In this section, we will explore the above ways of summarizing categorical data. We will also spend some time learning about tables as you will be using them extensively while working with categorical data. R has many packages for tabulating data and we list and explore all of them in the R scripts shared in the GitHub repository.

Click through for both guides. H/T R-Bloggers.

Comments closed