Launched: Data Science Portal


#21

@stanvx Congrats you found the first bug :slight_smile:


#22

I think what would calm everyone’s fears (and I had the same thoughts when I read the blog post) is just to include a line mentioning that non-hassio users can install the parts separately. Or even better, start off the sentence with that, and end it with “and for hassio users we’ve created this addon”. Just my 2 cents :slightly_smiling_face:


#23

i found that out when i did see your git rob.
the confusion comes because they lauched a doc that clearly states it expects hassio installed.
if i read your readme i see that it can be installed with pip also.

ir would have been very easy to make the docs so that its not only for hassio, but for all users.
and thats what worries me.


#24

Where do the docs say it is only for hass.io? It only assumes that people use hass.io in the Quick start guide and it does not say it can only be used that way.


#25

in the quick start it says:

This guide is assuming that you have a Hass.io installation up and running

from that moment on i dont know which part from the docs might be usable for other systems and which parts from the docs are specific for hassio.


#26

Well, the quick start guide is exactly that, a quick start guide. It is the fastest way to get you going. If you don’t use hass.io you usually are a more experienced user and know how to set up things like jupyter. But I agree that it should be made more excplicit that you don’t have to use hass.io.


#27

Where are the instructions for doing it not in hassio?


#28

Not is the docs right now. :wink:


#29

the whole quick start is for hassio users only. so how do i know if anything else in the docs is usefull for none hassio users?

if it is a home assistant doc it should at least have pointers to how to for others. off course an experienced user can start googling for jupyter, then find that site, find out how to install it and use it, and then find a way to use it with home assistant data.
an experienced home assistant user can do completely without the home assistant docs if needed. but he shouldnt!


#30

Ditto exactly what @ReneTode said :slight_smile:


#32

Would be glad to play with it and get some insights from it, running HASS using the official docker image


#33

How about a generic docker container? Please!?:grin:


#34

+1 - pls include also for non-hassio users


#35

I wish people would calm down instead of right away jumping to blaming the Home Assistant devs of destroying Home Assistant :man_facepalming:

It looks like it has been misunderstood, and so let me clarify: when we mention “data science portal”, we actually are talking about the new data website, which describes all the data that Home Assistant stores in detail and how you can query it.

Here is the definition of a portal by Google:

We have also made a JupyterLab notebook available that will help you explore your own data. Home Assistant itself does not contain a data science environment, so this notebook requires you to have JupyterLab installed.

For non Hass.io users, you are yourself responsible for installing applications on your system. You will need to figure out yourself how to install JupyterLab. After you do, you can use the same data science tools and notebooks as Hass.io users described in the quick start guide.


#36

Possibly you should wonder to yourself “Why does this keep happening?”, the answer may surprise you.

Not long ago this paragraph from you would have been “To do this on homeassistant, do XYZ. Hassio users can use an addon”. But nowadays it is “You’re not on hassio so work it out yourself”. This is the answer to what you should have been wondering.


#37

This doesn’t keep happening as you claim, so I am not wondering about that at all.

Please read my reply till the end. You can’t do this on Home Assistant, as it requires installing a standalone application that is not part of Home Assistant.


#38

I was just trying to help you understand why people are getting upset, but if you don’t think they are then that’s fine. It’s your project :slight_smile:


#39

Awesome. Looks pretty great, and almost no issues following the quick start guide.

The stumbling blocks I ran in to that could slow down installation:

  • JupyterLabs defaults to only binding to the loopback interface, and its installation documentation is a little light. It took me a bit more digging to find their docs on running a notebook server, which helped generate a config file and set it up to listen on all interfaces (locking it down again will be my next step :smiley: )
  • I had the same issue balloob has already highlighted in https://github.com/robmarkcole/HASS-data-detective/issues/89 . But running help(db_from_hass_config) gave me enough info to set the config path.
  • I’m using a MySQL database, and had to install extra modules before Jupyter could connect, which seems completely reasonable in hindsight :wink: . Including a link to the relevant SQLAlchemy docs listing supported dialects and the required modules might be helpful.

#40

i am sorry balloob, but isnt this exactly what i am saying?

you lauch a website for hassio users, tell none hassio users to figure things out themselves and from the website itself there is no way to tell which parts can be usefull for not hassio users.

i dont even want to use it, and i feel why this upsets people.
and it upsets people that you call your army, the ones that you need to make HA great and keep HA more alive then other programs.

and even if it is wrong that those people get upset, you should hear them why they are upset and see if it is possible to fix their feelings instead of just telling them to go figure it out themselves.


#41

The site says

That seems fairly clear cut.