Python DataScience JupyterLab on iPad/Android browser without Python Install -JupyterLite WASM-Based

Key Takeaways

[Music] dear fellow coders welcome to one little coder imagine a world where you don't have to have python installed on your machine but you still can actually open a jupyter notebook write some python code save it and then download it whenever you want you don't need python at all so imagine such a world and that world is a reality today with jupiter light jupiter light is a jupiter lab distribution that runs entirely in the browser built using jupiter lab components but the reason it is able to run entirely in the browser thanks to web assembly web assembly which is also known as wasp is nothing but a new type of code and it helps you run things on the web so instead of me trying to explain what is web assembly let me take you to an article that i read which i found good which is by argia so what is web assembly so basically web assembly is like a new uh low level code or it's not necessarily a code it is more like a target compilation code so what do i mean by that so you write code in a particular language c c plus plus java python or any of these languages and then you can convert this code or compile this code into wasp and this wasp is now run on a wasp a web assembly virtual machine and now this virtual machine can run can take this code which is actually written on a different language but compiled web assembly and run it in near native speed and it can run on any mission whether it is like your windows machine whether it is your mac machine whether it is like latest max m1 machine or even your arm devices so whatever the type of or final devices this wasp or web assembly lets you run the pre-compiled code in any place uh but yeah of course a browser so which means irrespective of whether you have a mobile device whether you have got a like something like a tablet or computer you can actually create near uh native speed uh applications without having to write on the same native uh language but instead of you having to still write a web assembly language you can just write it on your own language and compile it and then finally run it um that is enough for this video for you to understand what is web assembly but it is not just simply wave assembly enabling this there is actually a python stack that is compiled to web assembly so pio dyed is a is a python stack i should say or like a library so where uh you have got the python scientific stack like um scipy numpy pandas matplotlib so all these things are there and whatever this uh whatever code is written within pi pi or died is actually compiled to web assembly which means what does it mean it means you can actually write a python code uh especially like python data science machine learning code on the browser without having python installed so you don't have to have python installed at all but still you can write python code because ultimately what has happening is your python code using pyro data is getting converted into wasp wave assembly and then it is successfully run on the browser which means even though you are writing a code in a very high level programming language like python your code would work actually as fast as a native application on the web which is like you can say probably written uh with javascript or you can probably even say something like you know how assembly languages again uh one thing that i would like to highlight is webassembly is not here to replace javascript but to work very closely so that people who do not code in a typical web language like html css javascript they can actually write code on their own language and then uh compile it to wasp and then run it in near native speed so that is web assembly piodide is what is making jupiter light a reality now coming back to jupiter light what is jupiter like so jupiter light is nothing but a jupiter in jupiter uh hub uh where uh the back end the python kernel is backed by pyrodite so pyrodide uh ideally what is a kernel anytime you have a jupiter notebook you need a kernel which means it's like the language engine that does a compilation it's like a you know like uh if you want r then you need r kernel then if you want scala you'll need scalar kernel so these kernels make you make it possible for you to compile or run a particular language programming language and jupyter notebook so in this case jupiter lab pyro died is the engine the our python kernel is actually backed by pyrite and then there is a javascript kernel which is just an iframe because javascript is a native uh web language so it also enables you to store the notebooks locally and uh there are a couple of other features that you can check it out so what i'm going to show in this video is how we can just simply easily access jupyter lab and then we can write a python code and then ultimately of course um you can you can do some uh you know testing with yourself so let me get started so i will link this github repository in the youtube description and also fire droids link and web assembly please make sure you check out all these things and if you like jupiter light please give them a star and then appreciate them it would mean a lot to them so they've got two links one is jupiter lab the other one is like the retro lab which is like jupiter notebook let me start with jupiter lab so all you have to do is you have to click this thing and then it's going to open and then when it opens you can see how much time it takes it doesn't take a lot of time so you can see it's uh it's quite fast and then there are three kernels currently available you have bio light you have javascript kernel and you have got p5.js kernel so i don't know javascript so i'm going to simply get started with pyo light so new so i've got a new notebook you can see and i'm going to say import numpy as np import pandas as pd import plot lib dot plt uh as what am i doing pi plot as plt okay so to make things uh simpler and easier and you can see that it is getting executed uh you can see this uh icon here so it's getting executed and to make sync things simply like slightly simple i'm going to take a simple code to create linear regression linear regression and scikit learn okay let's uh let's see if we have any example data set so we have got linear regression cycle to learn so i'm going to copy these codes and then we are going to paste it and then we are going to run it so we have x and the shortcuts also work so i just did a control and y and then shortcut and you have linear regression and finally we have got the score or coefficient let's take the score and then let's what am i doing and then predict okay i don't want to intercept all those things okay i'm going to paste it so at this point you can see that we have got a working code of linear regression that uses scikit-learn and we have got a bunch of arrays where we are saying x y and then we want a linear regression so let's run one by one so first we are trying to import linear regression from scikit-learn and this is this is running you can see it it's executing while it is executing i would also like to highlight a fact that you can see there is a notebook that was created 35 minutes ago so there was a sample notebook that i created so which means uh using uh local storage on your browser so the notebook that you create uh will store it will get saved on your local browser so if you want you can download it locally so nobody is stopping you from the from doing that but even if you do not do that like you might be worried like what if what if like um you know a lot of times if there is a server which you do not have access to or maybe you don't get access to that thing but it is not the case here so something that you created before also still says so let's go ahead and then run the rest of the items and then you can see it just ran just like that and then we have got predicted output so i'm not going to explain you what's happening here so the idea is to show you that you can simply write a linear regression or let's say for that matter any machine learning algorithm or anything in python data science stack without having python installed on your local machine and even even one uh one greater thing is that you you're doing completely done browser so so it could be you know your ipad it could be your chromebook it could be your android tablet it could be anywhere or even a mobile phone where you've got only um only a browser so you can still do this thing so that is one thing the second thing that i wanted to highlight is uh something that i told you like i this is something that i created uh what am i doing okay the like i created when uh probably like 35 minutes ago but it is still accessible so what i'm going to show you is i'm going to quickly close this go back and then i'm going to open it again so that you know that whatever we created is automatically saved and that's what they're calling it support for save settings uh like offline notebook storage so i'm going to open jupyter lab again once i open jupyter lab again you can see basically it's actually a text sorry html file or what we are seeing the rendered html file so it's loading and you can see uh the file that we saved two minutes ago the file that we saved 37 minutes ago is available so they've got a welcome tour if you want to attend uh but other than that you know all the settings that you see everything is same so you can even export the notebook like if i have a notebook selected so let's say i have a notebook selected here and uh oops what did it do yeah i have a notebook selected here i won't export it i can go here click either download or export notebook as i can download it so um you may not be able to see that but my notebook is downloaded here so i just downloaded it so you can you can like leave it there then and there or you can even download the notebook and then the other advantage that you have is currently it supports only you know these three kernels but because it's an open source project because everybody is excited about wasp um it is highly possible that you would see support for a lot of other languages that's another thing that i wanted to highlight the other thing is it also supports uh multiple kernels running at the same time so for example like there are some uh jupiter servers where you can actually have only one kernel at the same time so like now i can have a python kernel open at the same time i can open a javascript kernel uh so you can see like i have a python kernel um but by this time you can see like there is a pilot kernel uh combined so now then i have a javascript console so let me just put two plus two is equal to four basically it's execute getting executed on javascript not on python but at the same time i can go here and then you know write some for uh i for i oops what am i doing for i in range of 10 print i okay it's it's an error because basically what i've done is i've written the code for yeah so you can see that uh it at the same time simultaneously you are able to use multiple languages but again now you might feel it is limited because you have got only three kernels um but i am very hopeful that uh the number of kernels would improve so you have a bunch of other options like uh you can see um you can see like what kernel it is you can shut down the kernel you have a theme options like you can change it to dark theme so these are basic options that you usually get with jupiter notebook uh and if you go to the project you would see a lot of other requests where you can uh like install extensions um like real-time collaboration is one of the latest features in jupiter lab which is i think it's still in development phase but yeah the like there are like active development you can see the latest comment is commit is yesterday that was done yesterday so there are a lot of a lot of good things coming out of it um and i am really hopeful about it but at least for now it is going to solve a huge problem where uh if you want to teach data science to somebody but like to be honest like ripple it is an amazing platform right why do people love ripple it people love reflect because like you if you want to start with a new programming language you don't have to necessarily go through the entire pain of setting up the environment installing something which is important but it is not important when you are starting with the programming language same thing with python data science like if you have to ask somebody to start with python data science they have to install anaconda they have to install uh they know how to they should know how to you know access kunda like some people may not like it some people may not like the smaller entry barrier uh but some for some people it is very important very important for them to get started faster than having a lot of uh brick walls uh and then you know finally them dropping off so now this is this is a very good example of where you don't have to do anything uh the in the environment side setting up the you know um your computation system but you can just click a link start with python data science and then also work it profi like use it professionally because uh jupiter notebook is is there everywhere uh like wherever whatever company you work for if the jupiter lab is there otherwise you know uh like systems like data breaks is there where uh that would also you know emulate jupyter notebook environment so you are in sure so all these things are there so please try it out i hope this is really useful for you and i hope you you really use uh it and if you are interested in wasp or if you are interested in development please uh try to contribute to this project um because uh this is an open source project you can see it is released in bsd3 license so you can see how you can contribute to it otherwise uh you know share your love for the developers and share the word with um with your friends and um and yeah like appreciate the developers on twitter linkedin wherever you go and again um you can use pyrodite without jupiter light that's another thing that i would like to highlight so you if you want to try just pyrite uh you can go here and then try pyrite but i think it makes a lot of sense to use pyrodine with jupiter lab uh like not uh not standalone so have a look at this also and then read more about web assembly i hope you would find it useful and interesting if this video was helpful to you please let me know in the comment section and if you have any uh feedback or suggestion for me let me know in the comment section otherwise please give a thumbs up subscribe to the channel if you have not and until next video stay safe