@carl, sounds like we overlap a lot! At least four of the main people on the HoloViz team spent 5-10 years in Scotland, I have family who lived for many years in Elgin, and several of us have been involved in fiber-optic sensing projects for oil wells using Panel…
I’am Nauman from Pakistan Islamabad. My initial work is in ISP in servers administration which later become passion for data exploration using AWK and Perl.
Then after joining in Jazz https://jazz.com.pk/ my basic job is to extract data from multivendor Network Servers and engineered it to for interactive data reports.
Using Emacs and Vim like tools i can reach and do scripting for data exploration efficiently
Started programming in python in 2015 converting network performance complex critical manual Excel reports to automated reports .
Then sharing my data exploration in Jupyter notebooks with my peers.
During 2020 Covid early work from home i started using Panel and Holoviews for automotive interactive dashboard application in comparison to my Tableau Dashboards.
Last year i started using Fastapi+Panel+Holoviews and i can say it is at par with Tableau Dashboards .
I really thanks the community in helping me out in my carrier paths
@jbednar just up the road then
That’s good to know about Panel and glad you have made some in roads there, I know two of the larger clients in the North Sea hav had smaller companies that were making use of plotly from what I could tell, it looked good but nothing I didn’t think panel and co can’t manage and sounds like to me it is. Actually the biggest disconnect in the field I find is having 3 or 4 people in the room at the same time when the tools are being devoloped. Somone that understands what is going on with the well, the programmer(s), someone that understands the data inside from a perspective of what is noise, why are there events going on & such and the end user. Maybe more but for me that would be the magic mix to making the best productive user friendly visual suite. All I know from an optics perspective is there is definately room for these tools to shine by bringing multiple data sources together and to make sense of it all in a visually appealling user friendly manner, this is the ticket.
@hyamanieu, that was an interesting video, thanks! Yann and I were co-presenters at a workshop on vision in 2012 when we were both still just professors; interesting that a decade later we are both managing open-source projects for corporations. Much of what he described also applies to my own journey turning arcane research code into tools powering real-world applications. So yes, maybe should get Peter Wang to interview me about that.
Thanks for tagging me! I’m an atmospheric scientist by academic degrees (a Bachelor’s and Master’s), but I consider myself more of a Python developer~ Currently, I am in between jobs, but soon I’ll be starting work at Prefect!
I was initially attracted to HoloViews because of the hover tool and how easy it was to step through timesteps to analyze a dataset, so I used it for my research! And before hvplot was released, it was kind of tedious to set labels and customize plots to make it “publication quality” so I went to develop HoloExt to make my own life easier, and even though it’s now deprecated (unusable), that was when I really learned how to use HoloViews more proficiently!
Eventually, after graduation, I continued using the HoloViz ecosystem a lot for building interactive dashboards at work (people were surprised by how efficiently I was able to create them, thanks to Panel). However, there were still things I couldn’t implement myself so I asked for a lot of help on this discourse. It didn’t feel proper to just ask but not answer others, so I decided to help others too, and I learned a lot through this as well!
Now, I am hoping Panel someday supports mobile app development because that would be super awesome!
Hi, I’m Nestor, an Argentinian researcher on electronic nanodevices (UNSAM - Universidad Nacional de San Martín) and distributed fiber optic systems (https://www.sur-tech.ar/, similar to what @carl commented, some version of the dashboard made with panel can be found here https://discourse.holoviz.org/t/distributed-fiber-optic-sensors-panel-multi-app-with-flask /1508), currently on a two-year sabbatical in Seoul. Throughout my PhD and my first years of research, I used Labview, C and Origin for work. Since mid 2019, when beginning to work with oil companies in distributed fiber optic systems, I started with python for signal analysis and alarm predictions. My collaborators used matplotlib, so a library was needed that could incorporate it to create a dashboard for the operators and supervisors of the oil pipelines that use the systems we developed. That’s how I found the panel library that allows you to combine bokeh with matplotlib, and thus we can use a lot of code already generated in the laboratory. In fact, to this day we still can’t get the same images of certain 1000 * 15000 matrices that plt.imshow provides, and we’re used to seeing it a certain way. The last thing I want to mention is the good community I found here, which helped me a lot in the first times with python and all the stuff, Mark, Philipp, etc… In fact, I still remembered my first PR with one line correction of a relative path for panel.
I hope you are all ok,
My name is Mike. I’m a clinical physicist working in radiation therapy. I spend much of my days working in our radiation oncology clinics but have always been interested in data visualization. I got into python during my undergraduate and graduate work and over the years have written a number of small tools and utilities for visualizing the large sums of data that are logged out by the linear accelerators we use to treat our patients. I’m currently the Chief of Radiation Physics for a hospital system in Colorado.
I have recently discovered Panel, Param, and Holoviews and am intrigued by the power of these tools compared to my “out of the box” python code. In an effort to learn the tools in my spare time I have started rewriting my old code specifically to utilize Param which has to be one of my new favorite libraries (I can’t imaging not having it at this point). I’m currently using these tools to develop a dashboard for visualizing trajectories for the radiation beam shaping components in the head of one of our particle accelerators.
I completed graduate work in a few different disciplines so to keep my mind going and my skills at least marketable I take consulting work in a variety of field and have found these visualization tools to be fantastic at just about everything.
In my spare time I like to enjoy hiking, camping, climbing and everything else the Rocky Mountain west has to offer. I really appreciate all of you and the help this group has provided as I have tried to develop the skills necessary to use these tools more effectively.
Welcome everyone and thanks for getting these introductions going! I originally set up this category but somehow managed not to actually write my own introduction…so here it is.
My name is Philipp Rüdiger, I grew up in Berlin but then pursued my academic career in the UK, first in York then in Edinburgh. After initially dabbling in music production (and electronic engineering) I ended up doing my PhD in computational neuroscience under @jbednar along with @jstevens. Initially I performed most of my analyses in MatLab but was quickly converted to Python by Jim and Jean-Luc and haven’t looked back since.
After Jean-Luc developed an initial prototype of HoloViews to visualize modeling output from the neural simulator I found myself needing to visualize not just the temporal evolution of the models but also analyze huge parameter searches. This led me to generalizing HoloViews to work with multi-dimensional parameter spaces and adding many different plot types. After that I spent most days and nights improving HoloViews and making little progress on my PhD. Eventually we were hired by Anaconda and I was able to implement a Bokeh backend and integrate HoloViews tightly with Datashader which had been in development at Anaconda. Oh and also eventually finish my PhD.
Through a consulting engagement with the UK MetOffice I then wrote GeoViews as a GIS extension for HoloViews. This was closely followed by another consulting engagement that allowed me to write hvPlot. It was initially envisioned simply as a plotting extension for the Intake data catalog library but soon took on a life of its own. After writing various prototypes to allow us to build dashboards from notebooks I was eventually given almost a full year in funding to go off and write a whole new solution which ended up becoming Panel and here we are.
These days I spend my days getting Lumen ready for a real release, maintaining and improving Panel and trying to keep up with at least some of the issues on the many other repos we maintain. Additionally I advise our clients on how to best leverage Panel for their use cases and use that experience to drive the development of new Panel features.
I’m really proud and excited to see the community around these tools grow, love seeing all your awesome showcases and most of all love seeing you answering each others questions on this Discourse. This is an exciting time for HoloViz and Panel in particular and as we expand our team we hope to continue growing this community and shipping new features, fixes and documentation more frequently.
Raghu here, a mechanical engineer who likes to code. Have implemented interactive visualizations utilizing Bokeh for engineering or scientific simulations. Lead an open source project for utilizing interactivity in STEM based classrooms. Happy to report it has at least touched 20-30 students in a classroom. Here are some of our sample work:
I have a use case for utilizing Panel and HoloViews and was planning to start a community thread as a Q&A to walk through the design and update progress. If this is discouraged, please let me know and I will post only if I hit some roadblocks and need help.
The reaction kinetics app looks fabulous! Please add that to the Showcase section. Yes, please do open a new topic for anything that might be of interest to the community or where you want feedback.
Welcome to the community!
As for the reaction kinetics app, have you thought of following what is shown in the right plot by adding e.g. a pointer on the curve of the left plot to show where we’re at?
If you wish, you can continue this conversation by opening a showcase topic as suggested by jbednar
Will open a new topic for future projects.
You can certainly double dip! Most of the examples on the Bokeh site won’t be HoloViz examples, and so for people looking for a HoloViz example, here is probably where they will find it!
Dear brothers and sisters,
I’ve just known this community from Marc. I am a mechanical worker, have very low experience in programing but I am very like concept of Panel. So I try to build my web app to leverage my products with Panel.
As new user, there are several things to learn. Please do not blame me if I have fool looking questions :).
My name is Ian Thomas, I am British and live in Caerphilly (famous for its castle and cheese) just north of Cardiff in the UK. Today is my first day working at Anaconda as part of the HoloViz team.
My background is physics and I have a PhD in ocean modelling. But I have mostly been a software developer for more years than I can count on my fingers and toes. I have worked in lots of different areas, so I know a small amount about lots of different things but not very much about anything specific!
I enjoy contributing to Open Source, mostly visualisation or more precisely mathematical algorithms that are used in visualisation. For Matplotlib I wrote most of the triangular grid work and the current quad grid contouring algorithm, and more recently I have become a core dev of Bokeh, mostly working on the WebGL renderer.
Through Bokeh I have been lucky enough to work with Jim Bednar and Philipp Rudiger on some improvements to Datashader, which is what brings me to Anaconda. Mostly my interests and skills are at the bottom of the HoloViz stack, i.e. Bokeh and Datashader, but I am looking forward to finding out more about HoloViews, Panel and hvPlot too.
Welcome, Ian! The bottom of the stack is definitely in need of some love!
Hello, I am Adrian, and I work as a Data Scientist at a small research lab. I typically work with clients who want to introduce ML into their manufacturing processes.
Holoviews is becoming my go-to visualization tool for EDA, communicating/reporting results to clients, and dashboards.
I have previously tried altair and plotly, but I think holoviews is the sweet spot between aesthetics, “declarativity” and “matplotlibness”.
My name is Ivan, from Italy. I’m unemployed at the moment . I spend much of my time learning new geo-python related stuff, trekking, growing vegetables (in summer) and preparing pizza , which I love so much!!!
I have a master’s degree in Geology and one in GIS. For 3 years, I’ve worked as a GIS tech in the utilities industry in the beautiful city of Siena (Italy, Tuscany). I loved this city so much, its history, the food and the amazing Tuscany landscapes .
During my job, I’ve mainly used Python and SQL to process and analyze spatial/non-spatial data togheter with QGIS.
Now, I’ve discovered Panel and… it’s powerfull!!!
As you’re a geologist who studied spatial data & an SQL user, how much are MULTIPOINT SQL types and the other spatial types used in the field?
I find it hard to use when encountering it and it always end up of doing a lot of translation work in python…
MULTI-type geometries are not that easy to manipulate. I haven’t encountered many MULTIPOINT types in my past job, only when I dealt with stratigraphic or postal data. Instead, I encountered a lot of data with MultiPolygon types, which is very common for data like administrative regions (regions with enclaves, islands etc), geology, land use, water bodies etc… Other geometric types I’ve encountered are LineStrings like road networks, utility networks (gas), river networks, etc…
For MULTI types I always used the ST_Dump in conjunction with ST_Collect and other ST operators. But most of the time, like you, I used Python (shapely and geopandas).