The problem with Jupyter isn't with what it does. It's the people who use it.
My experience as a data engineer/architect/application developer attached to data science teams for a while now is that most really good data scientists are very good at what they do, write somewhat competent code, and do not--in any way--care about writing good software or good application code.
Jupyter is a bane of my existence because people who use it want to use it for everything. Oh, it can have a web interface? Okay. The app is done. DEPLOY TO WEB USERS! NOW!!
It's a great tool. A lot of the people who use it are not software engineers, and they don't want to be. For a lot of people it's the straight line from point a to point b.
But in my experience, legit data scientists are pretty smart and are willing to learn a little if you're willing to give a little. This is a good exercise because they are typically skeptical about everything. So you have to be really secure about why you want certain things done certain ways, and why you definitely don't want things done other ways.
It's a good exercise for everyone involved if you have the right team dynamic and mutual, healthy respect for each other.
If you don't . . . well, then Jupyter notebooks completely suck.
My experience as a data engineer/architect/application developer attached to data science teams for a while now is that most really good data scientists are very good at what they do, write somewhat competent code, and do not--in any way--care about writing good software or good application code.
Jupyter is a bane of my existence because people who use it want to use it for everything. Oh, it can have a web interface? Okay. The app is done. DEPLOY TO WEB USERS! NOW!!
It's a great tool. A lot of the people who use it are not software engineers, and they don't want to be. For a lot of people it's the straight line from point a to point b.
But in my experience, legit data scientists are pretty smart and are willing to learn a little if you're willing to give a little. This is a good exercise because they are typically skeptical about everything. So you have to be really secure about why you want certain things done certain ways, and why you definitely don't want things done other ways.
It's a good exercise for everyone involved if you have the right team dynamic and mutual, healthy respect for each other.
If you don't . . . well, then Jupyter notebooks completely suck.