Dion Research LLC

an Analytics, Data Science and Visualization Company, founded 2011 Creators of @visu_ai
subscribe to our blog: or

[Dion Research LLC @DionResearchLLC]

We spend a lot of time in R&D, but we are more!

We have updated our logo to reflect how our products, like @visu_ai, are designed to help humans do their job better, through suggestions and explanations. The new logo suggests a brain and a text bubble.

#analytics #datascience

[Dion Research LLC @DionResearchLLC]

If you want to try ClassGraphic in a jupyter notebook, you can check it out through mybinder at the below link. This is a multiclass classification demo

#jupyternotebooks #python #datascience #plotly #classification #classgraphic #infoviz #dataviz

https://t.co/vDmVSDlwEi

[Dion Research LLC @DionResearchLLC]

The Dion Research ClassGraphic python module in action
#dataviz #classification #diagnostic

[Francois Dion [email protected] @f_dion]

The following data science plots are all static images, since that is all LinkedIn allows. But all these are web based interactive charts based on #plotly for #python. They can be used in #jupyternotebook, #pyqt, #pyside, #flask, #django or plain python s…https://t.co/6OZMRs1XpN

[Dion Research LLC @DionResearchLLC]

The documentation for #ClassGraphic (#python module), the #interactive #datascience #classification #diagnostic #metrics and #visualizations is at:

https://t.co/uVMiRoofsO

[Dion Research LLC @DionResearchLLC]

We just published the first public release (v0.2.0) of #classgraphic for #interactive #datascience #classification #diagnostic #metrics and #visualization

#dataviz #infoviz #ML #MachineLearning #Analytics #python #pandas #sklearn #plotly

https://t.co/KmmRAl0Vro

[Raymond Hettinger @raymondh]

When people first encounter @dataclass or NamedTuple in #Python, it is common for people to believe that the type annotations will generate type checks or type coercions.

@ dataclass
class Person:
name: str
age: int

Person('Alice', '30')

1/

[Dion Research LLC @DionResearchLLC]

@IgniteMindz The andromeda strain. Published in 1969, would become his first best seller. As it was written before that, he didnt have any pressure to deliver or to dumb science down & such.

Would also be made in a movie in 1971, using pre CGI techniques, such as projectors for holographs

[Francois Dion [email protected] @f_dion]

Made some progress on the bit crusher pedal prototype. I have the control knobs for the sample rate & the bit depth. Also have the overdrive stage. I am far from done, haven't added LED, bypass, & quite a few more things. It's already pretty effective on my Sequential drumtraks

[Francois Dion [email protected] @f_dion]

So how can you test this feature? Install Hotelling and dask and then in your #python script or notebook:

from hotelling.helpers import load_df
From hotelling.stats import hotelling_t2

x = load_df("pop_a_*.csv",dask=True)

Load y similarly then you can call:

hotelling_t2(x,y)

[Francois Dion [email protected] @f_dion]

My (@DionResearchLLC ) Hotelling #python package v0.5.0 now available! It's been beta tested for > month by companies to monitor #manufacturing #quality & large scale #financial #audits in another. Now supports very large datasets.
#stats #anomaly #pandas
https://t.co/5MavneO00H

[Dion Research LLC @DionResearchLLC]

Along with documentation, including example notebooks at:

https://t.co/armdy94Mrg

These notebook examples even cover the brand new deviation feature (cumulative sum chart).

#datascience #dataquality #dataviz #controlcharts

[Dion Research LLC @DionResearchLLC]

Hotelling v0.4.3 has just been released on pypi, now with plotly templates support and deviation charts (cusum)

https://t.co/POzJ3G1i6K

#runcharts #controlcharts #dataviz #infoviz #stats #t2 #tsquared #multivariate #matplotlib #plotly #Hotelling #deviation #qa #quality

[Dion Research LLC @DionResearchLLC]

While at it, not only did we update @stemgraphic documentation, but we've also pushed the latest documentation for Hotelling, the python package for T2 tests and uni/multivariate control (run) charts:

https://t.co/armdy94Mrg

#quality #datascience #eda #statistics

[Dion Research LLC @DionResearchLLC]

And to go along @stemgraphic 0.9.1 on https://t.co/DdMlDNIuaG, documentation has been updated at:
https://t.co/yJTcSkpM68

[stemgraphic @stemgraphic]

stemgraphic v0.9.1 has been released on pypi. From stem-and-leaf plots to stem-and-leaf heatmap grids, now working with the most recent #pandas #cufflinks #plotly

#EDA #DataVisualization #dataexploration #stemandleaf https://t.co/bQU2xEYMZP

[Francois Dion [email protected] @f_dion]

Another example of automated title on a fairly detailed plot, a scatter graph matrix. Title includes the dataframe name, the selected variables (without units, unlike the x and y axis), references the legend, the size that is not on the legend.

Good news? PR merged in #pandasGUi

[Dion Research LLC @DionResearchLLC]

@nicolaskruchten This one is not interactive, but one could easily see how clicking on the labels would show/ hide individual curves.

See scikit rf https://t.co/jluS1cE10j

We’ve (@f_dion) used those quite a bit i would imagine hover would give you the real and imaginary part.

[Nicolas Kruchten @nicolaskruchten]

📢 Plotly.js v2.0 is coming soon, after 5 years and 177 releases since v1.0! 🎉Check out the pre-release announcement to find out what's changing and how you can kick the tires on the Release Candidate 🚀 https://t.co/ePlYS18YaC

[Dion Research LLC @DionResearchLLC]

We use @visu_ai in all our #datascience, #dataquality and #datavisualization engagements.

[Francois Dion [email protected] @f_dion]

Excel works, until it doesn't. With scheduled checks, @visu_ai (https://t.co/TNSt47Stdj) would have caught this issue (spelled out, in this case), and many more. Data quality matters. Get in touch to learn more.

#datamanagement #d…https://t.co/kVk5eDmpIt https://t.co/Q8MO3yOgwp

[Francois Dion [email protected] @f_dion]

"Seeking Exotics"

#anomalydetection #outliers #inliers #ml #datascience

https://t.co/TYvr8WuC7O

[Francois Dion [email protected] @f_dion]

If you've started using Python 3.8:

- did you know about the walrus operator? ( := aka assignment expressions or pep 572)
- how often do you use it?
- what do you think of it?

I'm actually growing fond of it, after doubting it in…https://t.co/95GvqUeKBp https://t.co/SAr11DKIIT

[Francois Dion [email protected] @f_dion]

50 years ago, John Tukey introduced Exploratory Data Analysis. Imagine walking into his classroom in 1970 to witness history in the making: beyond his philosophy of EDA, he introduced several plots, including the box plot, one of the most popular statisti…https://t.co/1teOAx0kii