Launched: Data Science Portal


#42

that part is added a few hours ago AFTER people got upset.
edit: and its just what i said should have been there in the first place


#43

So? It’s there now. Your statement was incorrect.

And it’s there. Why is this still a concern?


#44

which part is incorrect? i didnt talk about how it is now but about the time it was lauched.

its still a concern because the reaction to the upset people was: “go figure it out yourselve” instead of:
“he we did see that we missed something, but we took care of it and added it to the docs”

my concern is that it will happen again if there is no awareness for that problem (that it upsets none hassio users if they feel left out)
and every time it happens again, the risk is that some great people from this community that keep the community alive, turn their back to the community.

and if balloob only reads from that, that i say that devs are killing HA, then he should read it again.
i say that it will kill HA if there is not enough concern for those who make HA great (and believe me allthough i am here on the forum daily and helping out in some areas, i am absolutely not talking about myself)


#46

Suggest to log issue here.


#47

Reported, Thanks


#48

Hi guys,

a possibly stupid question from my side: what is the advantage of this solution against InfluxDB+Grafana?
I see you do not need to duplicate your Database and rely on a single source (HA DB) but…I see a lot of advantages to keep a light and quick DB for HA (keeping like 7-10days of data recorded) and a separate DB for long terms statistics.

I am sure I am missing the point so can someone explain what are the practical advantages of this solution?


#49

Nice work done by the developers!

What i don’t understand, why the devs put a lot effort to create a Hassio addon, who’s targeted audience normally are the less ICT skilled people, to make home assistant and all its features accessible for those users. I have looked at this packages and come to the conclusion that this doesn’t fit the target audience of Hass IO, but more the people that have more ICT skills. I would be more logical to provide a more detailed guide to use these tools without Hass-io, so I understand the feeling of some users, but hey we are clever enough to figure it out by ourselves and then share it with this community. The tools that are provided here and the underlying python data analyses capabilities are tools that are used by professional datanalists/datascience people. The Jupyter notebooks are really great and reminds me of my earlier experience (20 yrs ago) I had with mathlab, where graphs/code/expressions could mixed up in documents. This is great for people who understand the content, but not so great for people who are just interested in the results (graphs). I must agree with @mspinolo that for the ordinary end-user, Grafana a more compelling interface is for Hass IO end users, who are interested in beautiful dashboards. The best of both worlds would be, do analytics and data mashups with python data tools and present the results on a Grafana dashboard. A lot of easier tasks/analytics can probably also be done in Grafana without python.


#50

With the add-on, an (aspiring) data scientist can get his data science tools ready to use with 1 click compared to having to figure out how it works on his system, how to connect to HA etc. What do you think will get more people to try it? What do you think will get more people actively developing new algorithms and contributing them back to HA? Right, it’s the one click that makes the tools available.

For the people in this thread, try re-reading all your comments and knowing that the people you’re attacking, volunteered their spare time and resources to built this. They spend their time to make a platform to allow data scientists to help with the development of Home Assistant. They didn’t get paid to do it, they did it because they believe in our mission of creating an independent home automation alternative that focuses on local control and privacy.


#51

i reread all the reactions 3 times more and i can only come to the conclusion that you feel attacked by me.
and i am sorry about that because that was not my intention.

all i did was try to speak out some concerns, just exactly because i have the respect for those who build HA in their spare time. and i hoped to help avoid upset people in the future.

so again sorry if some people feel attacked by my postings, that really was not my intention.
i hope you reread my postings without feeling attacked and see what i want to say.

i guess i can add nothing more then this excuse, so i leave it by that.


#52

Deploying Hassio today literally saved me hours of mucking about with Nginx, Duck DNS, lets encrypt and all that other nonsense. I was up and running on my otherwise bare metal Ubuntu 18.04 in no time. Docker is awesome, and so is Hassio. Thumbs up to the devs.


#54

From this prespective point and goal, to get aspiring datascientists involved, i now fully understand. Thanks for the clearification @balloob.


#55

Nobody is attacking anybody, people are asking valid questions. The reason they are so upset is because they are just as passionate about this project as you are.

The fact that some of us can’t write a new component doesn’t stop us dedicating hours and hours of our lives to this project in other ways.


#56

Wow the whining in this thread is unreal. When a new feature is released, why wouldn’t it make sense to design it (and the docs) first for a common install base?

Secondly, more than one person has said they think the focus on Hass.io for new features is disturbing or “concerning” and they imply that if the devs focus on hass.io the project is doomed. Seriously? Do you hear yourself? Hass.io is the very opposite. It ensures broad accessibility and I would wager has brought — and kept — more people in the fold than any other one thing. Anecdotally, I know dozens of people who wouldn’t have touched Hass before Hass.io — not because they’re dumb but because they don’t have the time to fiddle with things to the degree necessary without Hass.io.

I’ve done various install methods and have used HA since about version 0.15 I think. And I too now use Hass.io. It’s the future of this platform and it will enable devs to support the broadest possible user base and focus on higher level features rather than dealing with lots of low level environment-specific issues. Sorry if that means those of you running custom installs lag a bit on some features, or have to gasp wait a bit for more specific docs to guide you, but that’s just reality if you’re using a non-standard setup.

To @balloob, @frenck, @robmarkcole and others who work so hard on all this stuff…thank you. Your efforts are appreciated.


#57

that is never implied by anyone.

i think hassio is great for people who dont have skills or time to develope things theirselve. and it makes using it easy and thats great.
but i think that most of the people who use it are not the ones that develope new components, or repair components when something changes.

i might be wrong, but i think that the most people that use HA without hassio are the ones that are on the list of hundreds of people who contributed to HA.
we need those people.

its great that those people can use HA now also, but those are not the people we can rely on to expand and maintain HA in the future.

i can only second that.


#58

Really sad to see this thread, as well as the Twitter rant by Paulus. It sadly shows the project starts to fragment and fall apart.

I hope Home Assistant finds the stability in leadership it needs into 2019 and beyond.


#59

I’m an happy Hassio user and I’m excited by the idea to do some deep analytics (not only monitoring by Grafana) on my HA data. Unfortunately, from yesterday, I’ trying to install the add-on but I’m continuing to take the following error:

ERROR (SyncWorker_8) [hassio.docker.interface] Can’t install hassioaddons/jupyterlablite:0.1.0 -> 404 Client Error: Not Found (“no such image: hassioaddons/jupyterlablite:0.1.0: No such image: hassioaddons/jupyterlablite:0.1.0”).

I’m using Pine64 with Dietpi OS. HA 0.84.6 and Hassio supervisor is 141.

I really appreciate your help to fix the problem. Thank you very much in advance

ADDED INFO:
I tried to pull manually the image with the command:

  • docker pull hassioaddons/jupyterlablite:0.1.0

and I obtain the following message:

  • no matching manifest for unknown in the manifest list entries

#60

You’ll be pleased to know that this thread serves a purpose. It is being used by the Linus Torvalds School Of Management to demonstrate how to become tone-deaf to users’ concerns.

/s


#61

Well, it seems again that there are a lot of assumptions.

Let me clarify a couple of things about the new Data Science Portal. We provide Jupyter notebooks to interact with Home Assistant’s data for quite a while now. They can be used in multiple ways and give the users several possibilities to work with their data. The drawbacks of the standalone notebook approach was it didn’t work well with Hass.io or ARM-based device and required a lot of manual setup to get the data in an useful form.

The new add-on and HASS-Data-Detective are simplify the whole process and make the analysis of the data more approachable.


#62

Again? No no no, this has never happened before :laughing:


#63

I can’t believe that bullshit. In what way does that show the project falling apart? Releasing this as a addon for Hass.io makes things more accessible nothing more nothing less. It is just a quick way to get started. People who have a more advanced or custom setup already know how to set up those kind of things. This comment is nothing but ridiculous.