►
From YouTube: Education & Workforce WG: Analyzing the Short-Run Impact of C19 on the US Electricity Sector
Description
Date: 10/02/20
Presenter: Dr. Le Xie
Institution: Texas A&M University
Title: "A Cross-Domain Approach to Analyzing the Short-Run Impact of COVID-19 on the U.S. Electricity Sector"
http://sbdh-prod.ideas.gatech.edu/resources/newsblog/education-and-workforce-working-group
A
B
A
A
So
while
everyone
is
signing
in,
I
want
to
give
the
overview
of
the
working
group
again
for
those
that
may
be
new.
So
this
is
the
data,
science,
education
and
workforce
working
group.
This
is
a
group
for
its
open
to
professionals
and
data
data,
science,
educators
and
program
leaders
to
talk
and
hear
from
one
another
as
well
as
learn
about
resources
that
are
connecting
data
tools
and
industry
partners
in
research
around
this
area.
So
we
have
three
focuses
for
this
group
through
focus
areas.
A
So
I
welcome
all
of
you
here
and
we
have
a
goal:
hi
martin,
how
are
you
or
introducing
everyone
if
people
who
have
just
joined
can
also
look
at
the
chat
and
click
on
the
link
for
the
ether
pad
to
sign
in
and
also
see
the
agenda
for
today
and
take
notes?
So
as
we
go
through
all
right,
so
as
we
go
through
the
goal
overall
for
the
working
group
is
to
one
have
a
connected
network
of
data
science,
educators
and
program
leaders.
A
So
that's
you
so
we
hope,
through
these
ether
pads
that
we
get
to
connect
with
each
other
more
and
you
get
to
connect
among
each
other
more
to
learn
more
about
your
programs
and
what
you're
doing.
Second,
we.
Hopefully
we
are
going
to
be
taking
notes
through
the
dallas
big
data
hub
and
also
your
notes.
Looking
at
these
ether
pads
to
try
and
get
a
recommendations
guide
based
on
the
presentations
we
see
for
teaching
project-based
data
science
courses
and
hopefully
a
resource
list
as
well.
A
So
we
really
encourage
you
to
put
links
and
other
things
that
you
feel
would
be
helpful
to
the
community
or
that
you
would
like
to
see
in
the
ether
pad.
We
will
take
those
in
and
try
to
make
a
resource
list
and
also
models
for
evaluating
data,
science
courses
and
programs.
So
if
you
know
anyone,
that's
doing
this
or
you're
doing
it
or
you
have
a
thought
about
it,
then
please
put
that
in
so
for
that
so
I'll
wait
for
myself.
A
First,
I'm
dr
nada
rawlings-gus,
I'm
the
executive
director
of
the
south,
big
data,
innovation
hub.
It's
one
of
four
big
data,
innovation
hubs
that
are
serving
to
connect
the
data
science
ecosystem
across
academics,
industry
and
government
partners,
and
so
today
we're
going
to
have
two
speakers
that
are
going
to
be
talking
about
their
experience
with
their
data
science
project
and
first
up
is
going
to
be
dr
lee.
She
and
his
talk
is
going
to
be
on
a
cross-domain
approach
for
analyzing
covet-19
data,
but
using
the
electricity
in
sector.
A
It's
actually
looking
at
electricity
data
and
how
that
can
impact
covet
so
really
interesting.
So
he's
a
professor
in
the
department
of
electrical
engineering
and
he
is
at
the
university
of
texas
a
m
and
he
is
the
director
of
the
energy
digitization
at
texas,
a
m
energy
institute.
So
we'll
start
off.
If
you
can
share
your
screen
lee
and
we'll
start
today
and
as
we're
going,
if
you're
in
the
ether
pad,
you
can
ask
any
questions,
there's
a
section
for
questions
for
the
presenter.
B
Okay,
good
morning,
everybody-
and
it's
my
great
honor
and
pleasure
to
be
joining
this
south
big
data
hub
data,
science,
education
and
the
workforce
working
group,
thanks
ranatha
for
giving
us
the
opportunity
and
what
I'll
be
talking
about
in
next
20
minutes
or
25
minutes
or
so
is
some
of
our
recent
work.
That
is
focusing
on
using
a
data-driven
and
data
science
approach
to
understand
a
unique
sector,
namely
the
u.s
electricity
sector,
on
on
the
impact,
especially
the
short
run,
impact
due
to
covet
19..
B
So
I
guess
I
don't
need
to
motivate
any
more
about
the
impact
of
kovit.
Everybody
must
have
seen
the
news
this
morning
and
as
a
country,
we
are
far
from
you
know
being
over
this
this
this
challenge.
So
this
really
kind
of
started
out
just
as
the
country
went
through
the
lockdown
in
the
middle
of
march
and
all
the
students,
especially
those
grad
students,
postdocs,
were
staying
in
their
you
know
own
apartments
and
houses
and
so
on
and
we're
all
kind
of
just
isolated.
B
B
That
really
was
the
motivation,
and
you
know
I
should
really
give
a
big
shout
out
to
the
group
of
dedicated
graduate
students,
postdocs
and
my
collaborators
listed
here
for
for
making
this.
You
know
from
from
just
a
a
very
kind
of
a
quick
concept
at
the
middle
of
march
to
some
full
fruition
into
some
concrete
research
papers
and
and
projects,
and
so
on.
B
B
So
this
is
where
we
have
gone
through.
The
first
case
started
in
january
21st
on
march
13,
the
u.s
declares
the
national
emergency
and
practically
everybody
went
to
lockdown,
and
I
don't
know
about
you,
but
we
still
are
by
and
large
conducting
education
research
online,
even
as
of
today.
So
I
don't
even
I
lost
count
of
how
many
cases
and
and
so
on,
because
it's
too
many
and
it
looks
like
you
know
we
are
still
far
from
being
being
done
with
it.
B
So
just
kind
of
there
are
some
global
study
report,
notably
this
one
published
by
the
international
energy
administration
that
is
talking
about
sort
of
a
macro
level
impact
of
colgate
on
the
global
energy
sector.
The
kind
of
a
quick
takeaway
is
that,
because
of
the
economic
downturn
that
this
is,
you
know
bringing
about,
there's
likely
going
to
be
a
reduction
on
the
investment
on
the
new
power
equipment
and
resources,
so
that
is
going
to
have
some
impact
on
the
renewable
energies
continue
the
deployment
throughout
the
world.
B
This
is
something
that
we
kind
of
we
were
featured
in
one
of
the
recent
ieee
spectrum
articles.
So
I
want
you
to
just
take
a
picture.
Take
a
look
at
this.
You
know
beautiful
island
of
manhattan,
the
the
city
of
new
york,
and
this
is
something
that
nasa
capture
using
their
nighttime
image
data.
So
after
we,
of
course,
you
know
this
is
the
taken
during
the
midnight
of
february,
8th
and
april
25th.
B
Okay,
of
course,
there
are
some
ambient
lighting,
so
we
actually
took
some
algorithm
to
to
take
care
of
this
ambient
lighting.
So
what
you
see
here
is
pretty
much
the
the
lighting
of
the
human
activities
right,
electricity
lighting.
So
what
is
the
impression
you
know
the
average
brightness
of
the
picture
on
the
left
is
257
and
the
average
intensity
of
the
brightness
on
the
right
is
154..
B
I
was
literally
shocked
when
I
first
saw
it.
The
city
of
new
york
was
deemed
down
by
about
40
percent
and
what
is
causing
that,
and
you
know
what
it's
called
right,
so
it
has
actually
had
the
city
was
deemed
now,
and
you
know
that
in
a
very
visual
way
indicated
how
this
is
impacting
human
society
and
in
a
direct
way
on
how
it's
impacting
the
electricity
sector.
B
B
So,
just
to
kind
of
give
you
continue
to
give
some
of
the
visual
impression
how
does
covet
affect
human
society?
This
is
again
a
different.
You
know
colored
the
picture
of
the
one.
I
just
showed
you
about
new
york
city,
but
not
only
new
york,
pretty
much
all
the
major
metropolitan
areas
around
the
country
were
a
you
know,
substantially
affected
by
by
the
covet,
just
by
looking
at
the
nighttime
image
data,
so
houston,
boston,
l.a,
you
know
from
east
coast
to
the
west
coast.
B
And
then
this
is
how
the
data
hub
is
actually
being
architected
and
constructed.
So
you
can
see
there
are
a
few
pieces
of
key
data
sources
that
we
we
are
fetching
on
a
daily
basis.
Even
up
to
today,
one
is
across
the
nation.
There
are
wholesale
electricity
market
data
set
right
so,
for
instance,
new
york,
new
england,
texas,
california.
B
So
these
are
all
you
know,
electricity
market
at
wholesale
level
and
they
publish
a
lot
of
daily
data
on
on
electricity
sector,
but
we
didn't
stop
at
electricity
sector.
We
also
look
at
something
that
we
would
have
otherwise
never
touched,
for
instance,
the
public
health
data
set.
So
many
of
you
may
have
gone
to
this
john
hopkins
data
hub
for
kovit
19
related
information.
B
They
also
have
the
open
data
set
on
satellite
image
data
for
night
time
imaging
so
that
we
also
collect,
and
then
we
go
through
this
data
hub
of
you
know
some
automatic
data,
quality
checking
and
correction,
and
then
we
put
them
together
into
this
data
hub
and
you're-
welcome
to
check
it
out.
So
this
is
just
again
a
three
dimension
view
of
the
data
hub.
We
have
a
temporal
and
categorical
and
locational
differences,
and
just
for
those
who
are
not
necessarily
familiar
with
the
electricity
sector.
B
These
are
the
parts
of
the
u.s
that
is
covered
by
wholesale
level,
electricity
markets,
so
some
eighty-five
percent
of
the
population-
and
it
is
essentially
in
with
some
sort
of
a
wholesale
level
electricity
market
in
which
much
of
the
data
sets
are
becoming.
You
know
available
in
the
public
domain.
B
So
this
again,
you
know
one
of
the
natural
questions
for
any
kind
of
data.
Science
problems
is:
how
do
you
handle?
You
know,
data
qualities,
because
you
know
we
are
talking
of
massive
amount
of
data
coming
in
on
a
daily
basis
right,
so
you
have
covered
cases,
you
have
mobile
devices,
you
have
electricity.
B
How
do
you
make
sure
that
these
data
sets
are?
At
least
you
know
to
the
first
order?
You
know
had
some
level
of
data
quality
check,
so
we
actually
have
developed
a
automatic
process
to
figure
out
if
there
were
some
outlier
data
and
how
do
we
then
correct
them
using
historical
trends
right
so
just
to
kind
of
give
you
some,
then
some
deeper
quantitative
assessment
on
on
the
impact
on
electricity.
B
One
of
the
thing
that
people
care
a
lot
about
is,
is
you
know
that
does
covet
affect
renewable
generation
right,
because
the
renewable
generation
is
one
of
the
biggest
growing
sources
of
power
generation
in
the
electricity
sector
and
the
the
the
answer
is
not
much
not
much.
Perhaps
there
is
even
a
slight
increase
in
in
the
use
of
renewable
energy
just
because
of
the
the
general
trend
of
increasing
deployment
of
renewable
energy
resources.
B
What
about
the
prices?
So,
as
I
said,
there
is
a
85
percent
of
the
country
covered
by
wholesale
level.
Electricity
market,
the
market
will
definitely
have
prices.
In
fact
they
have
prices
for
day
ahead
and
for
five
minutes
ahead,
and
these
are
well,
you
know,
established
wholesale
level
marketplaces
because
of
the
load
change
electricity
demand
changes.
B
The
price
is
also
under
some
substantial
fluctuations,
they're
going
down
quite
significantly
in
most
of
the
markets
and
they're
going
down,
not
only
in
terms
of
their
average,
but
also
they're
sort
of
a
much
more
concentrated.
So
I
just
gave
you
a
example
here.
This
is
the
boston
hub
in
the
new
england
market,
so
you
can
see
the
prices
from
2017
to
2020,
going
leftwards
right,
leftwards,
meaning
the
price
on
average
is
going
down
and
also
it's
becoming
narrower,
narrower
meaning
the
concentration
is
there.
You
know
there
is
a
less
just.
B
Now,
let's
go
a
bit
even
further
on
on
the
more
kind
of
a
scientific
approaches
to
to
to
have
some
deeper
understanding
of
this.
You
know
what
we
have
presented.
You
so
far
is
very
much
a
kind
of
a
domain
specific
right.
We
look
at
renewable
production,
we
look
at
load
and
we
look
at
other
things,
but
we
felt
that
a
really,
I
think,
a
powerful
idea
and
perhaps
a
powerful
analysis
tool
is
to
really
go
beyond
a
particular
domain
of
data
set
right.
B
You
want
to
go,
maybe
cross-domain,
you
know
combining
electricity
with
public
health,
combining
electricity
with
mobility
and
so
on,
and
that
perhaps
will
give
you
some
further
insights
that
traditional
electricity
power
system
analysis
would
have
not
given
you
right.
So
that's
we're
going
to
talk
about
in
the
next
few
minutes,
and
this
is
really
has
just
been
featured
by
this
paper.
B
Just
you
know
fresh
off
of
the
press,
it's
going
to
it's
already
available
in
the
online,
but
it's
going
to
be
available
for
the
november
issue
of
the
jewel,
which
is
sort
of
one
of
the
you
know
most
prince
prestigious
journal
in
the
energy
sector,
so
the
details
are
all
describing
there,
and
so
one
thing
that
I
think
it's
actually
very
important
for
for
any
kind
of
data.
Science,
education
and
educators.
B
B
B
You
know
exclusively
on
the
impact
on
one
particular
event,
such
as
kobe
19..
So
I
guess
that's
what
economists
would
call
counterfactual
right
or
a
statistician
might
call
it
a
baseline
right.
So
so
I
think
it
is
important
when
we
are
analyzing
this
kind
of
impact
to
establish
a
counterfactual
or
baseline
meaning.
You
know
asking
the
following
question:
what
would
have
been
the
electricity
consumption
in
2020?
B
Should
there
were
no
covet
19
right
and
that
became
the
baseline
and
then
you
can
compare
the
actual
2020
electricity
consumption
with
that
baseline.
That
then
became
a
much
more
rigorous
attribution
of
the
difference
due
to
kobit
19
right.
So
I
hope
you
got
that
point
and
to
do
that
we
actually
utilized
a
a
a
deep
neural
network
approach
to
establish
this
backcast
model
and
that
turned
out
to
be
a
fairly
robust
and
rigorous
and
accurate
approaches
to
establishing
the
baseline.
B
Why
do
we
claim
so
so
perhaps
a
good
way
to
claiming
so
is
to
look
at
you
know
just
in
the
year
of
2020
right
before
kobe
hit
us
so
perhaps
looking
at
january
and
february
right
when
the
country
was
largely
not
affected
by
covet
yet,
and
then
you
look
at
this,
the
blue
curve,
which
is
the
real
electricity
consumption,
and
then
you
look
at
this
a
a
gray
curve,
which
is
the
counter
fracture.
You
see.
These
two
are
just
about
the
same
right.
B
B
You
can
see,
there's
a
drastic
difference,
so
I
want
you
to
focus
on
the
difference
between
the
blue
curve
and
the
gray
curve.
I
want
you
to
look
at
the
blue
curve
under
the
gray
curve.
Why?
Because
that's
the
real
difference
because
of
the
whole
bit
right.
You
should
not
look
at
the
difference
between
the
blue
curve
and
orange
curve.
B
Orange
curve
is
just
a
simple
looking
at
2019
and
the
contrast,
the
difference
between
2019
and
2020,
because
that
may
give
you
a
raw
impression
of
the
impact,
for
instance,
on
march
the
2nd
you
know,
while
the
impact
of
covert
is
still
relatively
mild,
you
do
see
a
difference
between
the
electricity
consumption
in
this
middle
curve,
and
that
was
largely
due
to
the
other
ambient
conditions
changes.
So
that
should
not
be
a
tribute
to
the
impact
of
kobe,
whereas
moving
down
to
say
deep
into
late
april,
where
new
york
was
really
heavily
affected.
B
B
Sure,
okay,
I'll
then
wrapping
up
my
talk
with
just
one
last
message,
which
is:
how
do
you
explain
that?
How
do
you
best
indicate
this
kind
of
changes?
So
that
is
really
again
this
cross
domain,
that
it's
going
to
give
us
a
lot
of
insight.
So
I
want
to
kind
of
give
you
some
visualization
of
the
mobility
data
of
the
u.s
population.
This
is
mainly
how
much
percentage
the
u.s
population
are
staying
at
home
and
that
the
more
towards
the
kind
of
a
pink
color,
the
more
we
are
staying
home.
B
B
So
I
just
want
to
leave
you
with
one.
You
know
technical
detail.
That
is
in
order
to
do
that
kind
of
analysis.
We
adopted
this
nobel
laureate,
professor
christopher
idea
of
establishing
this
vector,
auto
regression
as
a
way
of
understanding,
multiple
factors,
impact
and
on
on
the
value
of
interest.
For
instance,
we
are
interested
in
the
electricity
consumption,
but
we
have
a
number
of
factors,
for
instance
a
mobility,
public
health
and
staying
at
home,
and
so
on.
B
So
you
can
actually
develop
these
very
rigorous
tools
and
then
to
to
figure
out
what
exactly
is
the
most,
perhaps
highest
impact
factor
that
is
affecting
the
electricity
consumption
and
the
one
thing
I
want
to
highlight
you
is
this:
one
called
mobility
in
the
retail
sector.
It
turns
out
that
you
know
how
how
often
people
go
to
visit
the
retail
sector,
for
example,
how
often
you
know,
how
do
you
go
to
your
do,
your
grocery
shopping
and
and
so
on?
B
It's
actually
one
of
the
strongest
indicator
of
the
impact
on
the
electricity
consumption
changes,
so
so
that
again
enables
us
and
also
provided
us,
the
opportunity
to
conduct
more
rigorous
assessments
on
the
impact
on
on
on
the
of
covade
on
on
the
electricity
sector.
So
this
is
the
kind
of
the
framework
of
this
var
that
enables
us
to
do
such
analysis
and,
and
that
provides
insight
for
policymakers,
for
example,
in
the
city
of
new
york
and
instead
of
houston.
The
kind
of
impact
factor
is
actually
quite
different.
B
It
turns
out
that
in
houston,
you
know
the
the
visit
to
the
retail
sector
is
not
that
big
of
impact
factor
as
compared
to
new
york
city,
which
I
think
is
sort
of
interesting,
because
also
understandable,
because
new
york
city
is
perhaps
a
much
more
concentrated
commercial
area
as
compared
to
the
houston
area,
and
it
also
gave
you
some
hypothetical
questions
that
you
can
answer.
What?
If
scenarios
like?
What,
if
I
have
one
percent
increase
in
the
visit
to
the
retail
sector?
How
much
does
that
impact
electricity?
B
B
So
with
that,
I
want
to
just
kind
of
conclude
saying
that
we're
still
far
from
being
over
of
this
pandemic
and
we're
trying
our
best
to
see
what
we
can
help
as
a
public
service
and
education
on
helping
policy
makers
helping
planners
and
operators
to
kind
of
get
out
of
this
pandemic.
And
you
know
we
did
our
little
share
of
utilizing.
B
The
multiple
domains
of
data
sets
to
help
with
the
electricity
consumption
production,
which
is
actually
highly
correlated
with
the
social
economic
activities
in
in
general,
and
we're
pleased
that
nsf
has
just
given
us
a
new
grant.
That
would
enable
us
to
continue
looking
into
that,
and
with
that
I
will
stop
and
be
happy
to
discuss
later.
Thank
you.
A
Thank
you
lee.
I
I
know
we
were.
I
want
to
be
sure
that
we
have
deborah
on,
but
we
may
have
time.
I
had
one
question
about.
This
was
a
pretty
short
run
project
and
it
was
done
with
students
once
it
was
already
remote,
and
so
there's
a
lot
of
work
done.
How
did
you
coordinate
the
students
remotely
to
get
this
work
done?
B
Night,
well,
perhaps
more
seriously,
we
you
know,
I
I
think,
having
some
back-end
infrastructure
is
really
important.
For
instance,
this
data
hub
gave
us
a
lot
of
good.
You
know.
So
we
can
you
know,
so
we
establish
data
hub
precisely
to
make
sure
that
we
are.
We
are
collaborating
in
a
more
efficient
and
more
productive
manner.
B
We
also,
you
know,
conduct
a
very
frequent
discussions
over
zoom
and
you
know
I
don't
know
about
you,
but
we
kind
of
are
we
use
very
similar
to
what
you
have
done
for
this?
You
know
ether
pad
idea.
We
use
overleaf
right,
we
use
overleaf
to
share
and
collaborate
our
ideas
and
results
and
progress,
and
so
on.
So
I
I
would
say
you
know
I
was
quite
inspired.
B
You
know
by
this
process.
I
mean
just
like
everybody
we're
all
kind
of
locked
at
home,
and
you
know
we
feel
a
little
bit
kind
of
a
depressed
that
we
can't
go
anywhere
and
there's
nothing
but
much.
We
can
do
to
help
with
this
situation.
But
you
know
this
project
kind
of
gave
us
a
sense
of
purpose
and
and
and
mission,
and
it
kind
of
brought
us
together.
In
fact,
I
would
say
you
know
we
felt
that
we,
you
know,
even
though
physically
we
are,
you
know
separated.
A
Nice,
so
if
others
have
questions,
I
want
to
be
sure
we'll
have
might
have
some
time
at
the
end,
please
feel
free
to
put
them
in
the
ether
pad,
which
is
in
the
chat.
So
if
you
haven't
signed
in
to
the
pad
already,
please
do
and
then
also
add
any
questions
or
resources.
If
you
have
done
something
similar
with
the
project
that
you
want
to
ask
there,
we
will
be
looking
at
them
and
then
also
we
can
answer
questions
too.