For starters, Julia is not an actor or even a person. It’s a programming language and according to its creators, Julia fills a gap between easy prototyping and performance:
“The Julia programming language fills this role: it is a flexible dynamic language, appropriate for scientific and numerical computing, with performance comparable to traditional statically-typed languages.”
Hmmm. That’s interesting and while I have found many ways around Python’s performance (e.g., Azure), it’s exciting to think maybe I can train a model in less time on my workstation.
I really love Python. It’s easy, there’s an amazing ecosystem (via PyPi), and I…
I’m a competitive person by nature (or nurture, you pick) and I can’t help but wonder if I can beat PyCaret at machine learning. There’s really only one way to find out — have a competition!
My bias is already going to be almost overwhelming considering how badly I want to beat the machine, which means I need to establish some ground rules so it’s a least a fair fight (I imagine PyCaret is giggling to itself right about now). Here are the ground rules:
A few weeks ago, I created a YouTube video on connecting Microsoft Visual Studio Code to a Jupyter Notebook running on a Compute resource within Azure (via Azure ML Studio). When I was making the video, I kept wondering how much faster that remote server really is.
So I decided to test it.
Here’s the video for the Jupyter setup:
I’m not going to make you wade through my exploratory data analysis, or even loading my data. …
Implementing chatbots incrementally can not only save you the troubles associated with chatbots but can also service a valuable purpose to your organization along the way. One incredibly easy way to begin this journey is with Microsoft QnA Maker. (This article is not sponsored by Microsoft, nor have they had any editorial influence.)
It’s no secret chatbots are serving uses across industries, especially in customer service. They “talk” to customers, which is a decent self-service use case for chatbots, but it’s also a cautionary tale of “too much too soon.” …
Within the insurance industry, ACORD has been the standards body for normalizing information across the world. The origin of ACORD’s value to the insurance industry (back in 1970) was forms and back then they were paper. Now those forms are digital and that’s good news, but it can also be daunting because there are many versions of many forms (like, thousands of them). It would be nice if there was a service you could just call that knows about all those forms and versions and can just return data from an ACORD Form.
Regression modeling is fundamentally founded in predicting an outcome based on one or more predictors given a linear relationship between the variables. Wikipedia states it as
…linear regression is a linear approach to modelling the relationship between a scalar response and one or more explanatory variables.
There are some assumptions involved in linear regression, which I’m going to sidestep (see this link for more information) in order to spend more time on something that is unexpectedly (for me) both interesting and useful — creating a linear regression model using Microsoft Excel. …
Yes, I do have a ‘favorite’ line of Python code because there happens to be something I do so often (both in my word and academic lives) that the elegance of it not only saves me time, but also does what I need.
Pandas DataFrames are an especially good way to manage tabular data in Python. Often, in data science, data is manipulated — not just one or two values, but every value in one or more columns. For example, when making corrections.
That’s where my favorite line of code comes in.
Here’s an example with some bad data:
If you’ve working with the Azure Machine Learning Python libraries in Visual Studio Code, you probably started with a Microsoft tutorial, which showed code like the following:
from azureml.core import Workspace
Upon seeing that seemingly innocuous code, you (at some point) install azureml.core from pypi using
pip install like you’re supposed to. Paraphrasing the command line output, you end up with something like the following.
pip install azureml.core
<....a bunch of stuff gets collected...>
Installing collected packages: azureml.core
Successfully installed azureml.core
You’re ready to (re)try your import, not realizing you’re about to fall into an abyss. What comes next doesn’t…