►
From YouTube: JupyterLab Bifrost
Description
During the Cal Poly San Luis Obispo (SLO) Project Jupyter Summer Internship, a team of mentors from various partner institutions and companies trained a cohort of 10 students who built three innovative JupyterLab extensions that we’re excited to share with the community!
In this video you will learn more about the third extension:
JupyterLab Bifrost built by Angela Chawla, Jay Ahn, and John Waidhofer, mentored by Arvind Satyanarayan from MIT, Piyush Jain from Amazon Web Services, and Brian Granger from Amazon Web Services/Cal Poly.
Jupyter Bifrost is an interactive data visualization extension that allows users to edit charts and export updated pandas dataframes. It circumvents the tedium of static visualization and complexity of code. This allows users to focus on data exploration and quickly extract insights from datasets. https://github.com/jupytercalpoly/Jupyter-Bifrost
View the other two presentations, JupyterLab Commenting & JupyterLab Notifications: https://youtu.be/1ift0Ew1J20
A special thank you to Zach Sailer from Apple, Steven Silvester from Apple, and Kevin Jahns from Yjs for supporting and mentoring the entire cohort this summer.
Background on the program:
For more than 15 years undergraduate and graduate student interns have played a critical role in the Jupyter community. Over the years at Cal Poly SLO, 51 engineering and design students have built and helped to build some amazing things: IPython Widgets, nbconvert, JupyterLab, JupyterLab Table of Contents, JupyterLab Git extension, JupyterLab Status Bar, and others. Multiple current and former Jupyter Steering Council members and core developers started out working with us as students.
This program would not be possible without our funders: Leona M and Harry B Helmsley Charitable Trust, Alfred P Sloan Foundation, Schmidt Futures, Gordon and Betty Moore Foundation.
00:00 Introduction
05:10 JupyterLab BiFrost
23:30 BiFrost Questions
A
Hello
and
welcome
to
the
2021
capstone
event
for
the
project:
jupiter
internship
at
california,
polytechnic
state
university
in
san
luis
obispo,
that's
a
really
long
name,
and
so
we
call
it
cal
poly.
For
short,
I'm
thrilled
to
be
presenting
today,
I'm
thrilled
to
be
introducing
today
the
phenomenal
work
undertaken
by
this
summer's
cohort.
A
This
summer,
more
than
ever,
our
internship
was
made
possible
by
a
wonderful
group
of
community
mentors
and
collaborators,
but
I
did
want
to
take
a
moment
here
at
the
beginning
to
introduce
the
cal
poly
team.
We
have
brian
granger
founder
of
the
program
christine
moon
operations
coordinator
for
our
team
and
myself
anna
rivolcaba,
director
of
the
program.
A
I
also
next
slide,
please.
I
also
want
to
take
a
moment
to
acknowledge
the
funders.
Without
them,
this
program
would
not
be
possible.
This
list
includes
past
grants
and
current
funding.
For
our
group.
We
have
the
leona
m
and
harry
p,
helmsley
charitable
trust,
the
alfred
p,
sloan
foundation,
schmidt
futures
and
also
the
gordon
and
betty
moore
foundation.
A
B
B
The
first
iteration,
was
all
the
way
back
in
2005
when
I
was
at
santa
clara
university,
and
the
goal
here
is
to
have
students
solve
real
world
problems
that
matter
to
jupiter
users
and
includes
both
design,
ux
design
and
software
engineering,
to
build
something
and
ship
it
in
one
summer.
So
next
slide.
B
One
dimension
that
makes
this
program
unique
is
that
we
hire
students
in
teams
of
three
where
each
team
has
two
software
engineers
and
one
ux
designer
on
the
team,
and
we
really
emphasize
and
follow
a
human-centered
design
process
for
the
the
work
that
the
teams
are
doing
this
year.
One
thing
that's
new
is
that
we've
had
broader
members
of
the
jupiter
community
and
adjacent
communities,
mentor
the
interns,
and
so
at
this
point,
we'd
like
to
acknowledge
interns
for
this
team
next
slide.
B
So
for
this
bifrost
extension,
we've
had
arvin
satyanaram
from
mit
pierce,
jane
from
aws
as
well
as
myself,
and
then
we've
also
had
three
other
people
in
in
the
jupiter
community
mentor
students
available
to
to
answer
questions
zack
sailor
who's,
a
software
engineer
at
apple,
stephen
sylvester,
also
a
software
engineer
at
apple
and
then
kevin
johns
who's.
The
creator
of
yjs
fantastic
crdt
framework
that
we
using
in
jupiter
lab
now
who's
been
working
with
quant
stack
and
the
jupiter
lab
team
on
real-time
collaboration
in
jupiter
next
slide.
A
I
do
have
a
few
logistical
reminders
before
the
team
gets
started,
but
go
ahead
and
get
ready
to
start
now.
I
did
want
to
share
the.
We
have
a
couple
ways
to
communicate
with
us.
One
of
them
is
the
chat
channel
for
general
comments
and
or
any
issues
you
might
be
having
with
the
webinar.
The
other
is
the
q
a
section,
and
you
can
find
that
at
the
bottom
of
the
zoom
window,
where
you
can
enter
questions
you
can
you
don't
have
to
wait
until
the
end
of
the
presentation
to
enter
questions?
A
C
D
So
why
bifrost?
Let's
start
with
the
pros
and
cons
of
the
current
visualization
tools
used
in
jupiter
environment
and
discuss
what
makes
bifrost
so
powerful
visualization
tools
are
essential
to
summarize
and
understand
the
data
and
both
the
existing
and
existing
static
and
interactive
tools
serve
as
a
great
standard.
D
D
Second,
to
create
a
graph
to
create
a
graph.
It
is
inevitable
to
explore
data
first,
since
you
need
to
specify
columns
for
each
axis
and
mark
type.
You
might
think
this
could
not
be
a
big
of
a
deal
with
the
small
data
set,
but
with
the
large
data
set,
it
could
slow
down
the
process
of
data
exploration
and
quickly
extracting
key
insights
from
the
data.
D
D
To
sum
up
with,
all
of
these
wonder,
features
that
fiber
frost
provides.
It
really
helps
users
to
focus
on
extracting
key
insights
from
the
data
without
extracting,
without
getting
distracted
from
writing
a
code
with
that,
I
will
pass
on
to
angela
to
discuss
about
the
process
of
the
development
throughout
the
summer.
C
Great
thanks
jay,
we
started
with
research
first
into
the
visualization
tools
that
are
already
out
there.
Specifically,
we
took
inspiration
from
b2
and
voyager
data
voyager
from
b2.
We
took
inspiration
from
their
history
log
that
kept
track
of
interactions
with
the
user
interface
and
from
datavoyager.
We
took
inspiration
from
their
chart
recommendation
system
and
their
unique
system
for
defining
encodings.
C
C
C
C
We
then
experimented
with
a
branching
tree
history
system,
where,
when
users
went
back
in
time,
viewed
a
previous
date
and
made
new
edits,
those
edits
would
be
saved
alongside
their
previous
version
of
edits.
Originally,
this
was
conceived
of
as
a
very
literal
branching
tree.
It
was
later
simplified
to
a
file
system
like
metaphor
where
that
new
branch
of
edits
was
nested
underneath
of
the
state
that
it
was
based
on.
C
Next,
we
moved
on
to
user
tests.
Our
first
ones
were
structured
and
task-based
focused
on
testing,
ux,
mechanics
and
the
overall
intuitiveness
of
the
interface.
Next,
we
moved
on
to
open-ended
user
tests,
giving
users
a
data
set
and
asking
them
to
pull
insights
out
of
it.
This
helps
us
get
an
idea
of
the
natural
data
exploration
flow
of
bifrost,
all
right
I'll,
go
ahead
and
pass
it
off
to
john
for
a
demo.
E
Today,
we're
going
to
be
looking
at
a
california
fires,
data
set
from
2013
to
2019..
It
looks
at
the
destruction
caused
by
the
fires
where
they
occurred
and
the
resources
allotted
to
each
one.
We
want
to
see
if
there
is
any
connection
between
the
resources
allotted
and
the
destruction
cost
immediately.
When
you
launch
the
chart,
we
get
presented
with
this
column
screen,
and
this
is
a
really
nice
way
to
introduce
yourself
to
your
data
set.
E
E
E
This
chart
actually
seems
like
a
good
place
to
start,
because
it's
a
bar
chart
with
the
mean
of
acres
burned
by
active
beer,
and
now
we
can
go
on
and
edit
the
chart
looks
like
2013
through
2017
had
pretty
expected
fire
years
in
terms
of
acreage
burn.
The
2018
looks
to
be
an
outlier,
so
let's
zoom
in
on
that
and
go
up
here
and
brush
over
2018.,
and
I
want
to
subdivide
2018
by
the
fire
names.
E
E
E
E
E
But
this
isn't
necessarily
the
complete
story.
Crews
involved
are
definitely
the
core
of
the
fire
team,
but
there's
also
tankers
and
doors
and
engines
that
actually
either
move
around
earth
to
create
fire
roads
or
supply
water.
So
it
would
be
nice
to
have
an
overall
resources
cabin
to
do
this.
Let's
look
at
our
fields,
see
if
there's
a
resources
field.
E
Well,
it
doesn't
seem
like
there
is
so,
let's
jump
back
out
of
bifrost
into
good
old
panda's
data
frames
and
make
one
so
here
and
make
a
resources
field.
That
is
just
roughly
a
sum
of
all
of
the
resources
available
in
terms
of
engines.
Crews
involved
in
dozers
run
this,
and
before
I
plot
I
could
use
the
python
api
to
go
back
to
something
very
specific
and
skip
the
onboarding
section.
I
know
exactly
what
I'm
looking
for.
I
want
to
plot
the
resources
by
each
county
to
understand
how
they're
divided
throughout
the
state.
E
E
E
As
you
can
see,
it
seems
like
the
areas
with
a
lot
of
resource
have
much
fewer
burned
acreage,
which
would
make
sense,
and
then
areas
with
less
resources
seem
to
have
more,
and
if
you
want
to
easily
jump
between,
you
could
always
use
the
history
top,
but
the
lake,
the
lake
county
seems
to
be
an
outlier
here.
It
had
a
lot
of
resources,
but
it
also
has
a
lot
of
acres
burned.
E
This
is
something
that's
super
powerful
about
bifrost.
It
gives
you
the
ability
to
come
up
with
certain
to
see
correlations
that
you
went
before.
We
were
looking
at
counties
and
fires
as
like
a
one-to-one
relationship,
but
the
fires
themselves
don't
know
bay,
county
wants,
and
so
by
looking
at
this
graph,
we
can
clearly
see
that
these
two
counties
are
dealing
with
the
same
fire,
and
maybe
we
could
approach
this
data
exploration
in
the
future
with
a
more
holistic
view
throughout
the
state
instead
of
dividing
resources
by
counties.
E
Now
it
looks
like
this
ranch.
Fire
is
really
interesting,
as
well
as
the
campfire
in
butte
county,
and
I
want
to
see
this
why
san
diego
has
so
few
acres
burned
compared
to
the
total
number
of
resources
that
they're
allotted,
so
I'm
actually
going
to
isolate
san
diego
versus
butte
here,
and
I
want
to
look
it
into
it
and
pandas
in
any
other
program.
This
would
be
very
difficult.
You
would
have
to
take
what
you've
done
in
the
program.
E
Remember
everything
you've
done
in
the
graphical
user
interface
here
in
the
chart
and
then
recreate
it
in
pandas.
But
the
amazing
thing
about
bifrost
is:
I
could
actually
take
all
my
edits
and
put
it
back
into
pandas
and
just
keep
going
in
the
code
so
to
isolate
these
two
counties.
All
I
have
to
do
is
go
here
and
select
each
one
individually,
so
I'll
select
view
and
then
I'll
go
down
and
select
san
diego.
E
E
C
You
might
also
see
a
simplified
and
improved
user
interface
to
let
users
focus
on
exploration
of
data
instead
of
navigation.
Finally,
you
might
see
better
handling
of
large
data
sets
and
different
data
types
like
time
series
all
right.
Thank
you
so
much.
We
have
a
little
bit
of
time
for
comments
and
questions
now.
C
E
E
If
it,
if
it
isn't
able
to
be
stored
in
memory,
we
can't
necessarily
work
with
it
within
our
current
bounds,
but
we
do
have
the
ability
to
sample
these
large
data
sets
and
provide
a
subset
of
useful
data
that
is
representative
of
the
large
dataset
and
that's
the
approach
we
took.
So
there
is
a
sampling
feature
within
bifrost
that
allows
you
to
extract
insights
without
necessarily
looking
at
the
entire
data
set.
E
When
you
switch
screen,
we
up
sample
and
there's
actually
a
sample
slider
on
I'll
share
my
screen,
so
we
can
see
there's
a
sampling
slider
on
this
chart
screen,
and
so
you
could
actually
control
how
much
of
the
data
set.
You're
sampling.
You
can
see
the
number
of
points
that
you're
sampling
right
here,
and
this
allows
you
to
easily
sample
your
data
set
without
necessarily
having
to
go
through
the
work
of
doing
it
manually
in
pandas
and
whenever
you
jump
back
to
a
screen.
E
A
Right,
it
doesn't
look
like
there
are
any
more
questions,
so
I
just
want
to
say
big
congratulations
to
the
entire
team,
so
great
work,
everyone
and
also
to
the
mentors.
Thank
you
to
you
and
arvind.
This
is
an
excellent
presentation.
It's
really
fun
for
me
to
watch
the
the
culmination
of
your
summer's
work
and
I
also
want
to
provide
in
the
chat
link
to
the
jupiter
cal
poly
org
on
github.
A
You
can
see
the
entire
summer's
repos
in
there
one
more
resource
I
want
to
share,
and
this
is
a
general.
This
is
a
general
project,
jupiter
resource
and
that's
our
discourse
channel.
That
is
a
an
important
port
point
of
connection.
I
don't
feel
like
everyone's
aware
of
it,
so
I'm
taking
just
a
moment
to
acknowledge
it
and
share
it
with
you
all
and
so.