00:00:00
hey guys andy here with cqe academy and
00:00:02
in today's video i want to talk about a
00:00:04
really important topic
00:00:05
which is the seven qc tools now whether
00:00:08
you just want to get better at work and
00:00:09
use these tools in your everyday job
00:00:11
or you're preparing for something like
00:00:13
the green belt exam or the black belt
00:00:14
exam or the cqe exam
00:00:16
today's lecture is for you all right
00:00:18
let's head over to the computer get
00:00:19
started
00:00:20
all right let's go ahead and jump in
00:00:22
right into the agenda so we're gonna
00:00:23
start with a brief
00:00:24
intro of the seven qc tools kind of talk
00:00:26
about all of them and how they fit into
00:00:28
the
00:00:28
problem solving process or the the
00:00:30
improvement process and then we're gonna
00:00:31
go through each one we're gonna start
00:00:32
the flow chart
00:00:33
the check sheet the pareto chart the
00:00:35
cause and effect diagram
00:00:37
scatter diagram histogram and then the
00:00:38
control charts and then
00:00:40
along the way as we go through this
00:00:42
we're actually gonna work a problem
00:00:44
using all seven tools and we're gonna
00:00:46
reduce the number of defects
00:00:48
associated with our toaster all right
00:00:49
let's go and get started so the seven qc
00:00:51
tools
00:00:52
i love this quote from kerou ishikawa
00:00:54
who said
00:00:55
as much as 95 of quality problems can be
00:00:58
solved with seven fundamental tools and
00:01:01
i
00:01:01
absolutely agree with that i think these
00:01:03
tools are probably
00:01:04
the seven most powerful tools whether
00:01:08
you're talking about
00:01:08
green belt or black belt or quality
00:01:10
engineering it doesn't matter
00:01:12
these seven tools are incredibly
00:01:14
powerful for
00:01:15
solving problems and making improvements
00:01:18
and and this is a really important topic
00:01:19
by the way as we go through this i'll
00:01:21
make sure to talk about
00:01:22
where we're at in something like the
00:01:24
plan do check act or the domain cycle
00:01:26
we're gonna solve a problem with our
00:01:28
toaster and we'll we'll use either the
00:01:30
domain or the plan do check out process
00:01:32
to do it
00:01:33
all right let's get into it all right so
00:01:34
the very first tool is the flow chart
00:01:37
and what a flowchart does is say it's a
00:01:39
visual tool
00:01:40
that helps you depict the flow or the
00:01:42
sequence of a process
00:01:44
this could be things like the flow of
00:01:45
information or the flow of tasks or
00:01:47
material or people or decisions
00:01:49
it doesn't matter the reason that a
00:01:51
flowchart is so incredibly valuable is
00:01:53
it makes a
00:01:54
really complex process simple and
00:01:57
it promotes a common understanding of a
00:01:59
process anytime you get more than one
00:02:01
person in a room to talk about a process
00:02:04
there's likely going to be disagreement
00:02:05
about how the process works
00:02:07
and i love using this analogy often
00:02:10
times when when we sit down to
00:02:11
analyze a process there's what
00:02:13
management thinks is happening
00:02:14
there's what the procedure says is
00:02:16
happening there's what's actually
00:02:18
happening on the production floor
00:02:19
and then there's what could be happening
00:02:21
and the beauty of a flow chart is it it
00:02:23
does just that it gets everyone on the
00:02:25
same page
00:02:26
about what's actually happening and i
00:02:28
love this quote
00:02:29
from dr deming who said if you can't
00:02:31
describe what you're doing as a process
00:02:34
you don't know what you're doing and the
00:02:36
best way
00:02:37
to describe what you're doing is to use
00:02:39
a flowchart
00:02:40
and that's why this tool is so powerful
00:02:41
if you're in the planning phase of the
00:02:43
define phase
00:02:44
it's really good to use a flow chart
00:02:46
define your process
00:02:48
and then use that flow chart to plan out
00:02:50
your experiment and plan out how you're
00:02:51
going to make an improvement
00:02:53
so let's do just that let's say we're
00:02:55
talking about our toaster
00:02:56
and we want to make an improvement right
00:02:59
and so the first thing we're going to do
00:03:00
is we're going to start with the
00:03:01
boundaries we want to analyze a process
00:03:03
but we want to start with our boundaries
00:03:04
first
00:03:05
so we're going to go from receiving a
00:03:06
work order to completing a workload
00:03:07
that's the boundaries of our flow chart
00:03:09
now i've got the team here because all
00:03:11
of these activities all these tools are
00:03:13
all team based so imagine you're sitting
00:03:14
down with your team
00:03:15
and the first thing you're going to do
00:03:16
is brainstorm all of the steps in the
00:03:19
process right
00:03:20
talk to the experts how does the process
00:03:22
work use post-it notes right don't try
00:03:24
to do this in some software
00:03:26
use post-it notes write down all the
00:03:27
activities and then once you're done
00:03:29
brainstorming
00:03:30
organize those thoughts into that
00:03:32
logical flow
00:03:33
that logical sequence of activities for
00:03:35
your process
00:03:37
and now that we have our process here
00:03:38
we're in that planning phase and we want
00:03:40
to create a target right
00:03:41
what sort of improvement are we going to
00:03:42
make and we want to reduce defects by 25
00:03:46
now we can't make an improvement and we
00:03:48
can't solve a problem without data
00:03:50
and we know that most of our defects
00:03:52
happen during final testing
00:03:54
so now we need to collect a little bit
00:03:55
of data and this is where the check
00:03:57
sheet comes into play
00:03:58
so the check sheet is a very simple tool
00:04:01
for collecting
00:04:02
organizing and analyzing data every
00:04:05
problem you solve or every improvement
00:04:06
you make
00:04:07
should be based on data and the check
00:04:09
sheet is probably the most powerful tool
00:04:11
for collecting data
00:04:12
now there's something wrong with the
00:04:14
check sheet that i'm showing you here on
00:04:15
the screen
00:04:16
and that problem is is it doesn't have
00:04:17
any metadata
00:04:19
if you're collecting data and you want
00:04:20
to make a high quality decision
00:04:22
using that data you also need metadata
00:04:24
so when you're creating your check sheet
00:04:26
don't forget to include things like who
00:04:28
and when and where
00:04:30
all those key elements of data integrity
00:04:33
and data accuracy
00:04:34
are really important for making high
00:04:36
quality decisions
00:04:37
okay so we've got the team together and
00:04:39
again we did a little bit more
00:04:40
brainstorming we said
00:04:41
okay at final testing we have eight
00:04:43
defects that we want to collect some
00:04:44
data on so we create this check sheet
00:04:46
we've got our metadata here
00:04:47
we hand this off to the team and they
00:04:49
come back to us a week later
00:04:51
with a bunch of data now this is
00:04:53
fantastic we finally have some data
00:04:55
that we can analyze and the question is
00:04:57
which defect
00:04:58
do we focus on i want to improve our
00:05:01
target so we originally said
00:05:02
we want to reduce defects by 25 percent
00:05:05
well now that we have a little bit of
00:05:06
data we can actually create a target
00:05:08
so we have 145 defects across a whole
00:05:10
week that's seven days
00:05:12
that means we're averaging about 20 to
00:05:14
21 defects
00:05:16
per day now if we can reduce that by 25
00:05:19
percent
00:05:20
we will eliminate five defects per day
00:05:23
now we obviously can't focus on all
00:05:25
these defects so the real question is
00:05:27
how do we know
00:05:28
what to focus on and that's where the
00:05:30
pareto chart comes into play
00:05:32
so the pareto chart is another qc tool
00:05:34
that allows you to
00:05:36
analyze your data in search of the
00:05:38
pareto principle
00:05:39
so what it what is the pareto rule what
00:05:41
is this 80 20 rule
00:05:42
so this this is a a natural phenomenon
00:05:45
that was discovered by a guy named
00:05:46
vilfredo paredo
00:05:48
he's an italian researcher who was
00:05:50
studying
00:05:51
land ownership and wealth distribution
00:05:53
in italy
00:05:54
and in europe and what he found is that
00:05:56
80 percent of the land
00:05:57
was owned by 20 percent of the people
00:06:00
and this 80 20 rule in this 80 20
00:06:02
phenomenon
00:06:03
was also experienced by a guy named
00:06:05
joseph duran
00:06:06
now he gave credit for the tool to
00:06:09
wilfredo praeto but he was the one who
00:06:10
popularized this idea
00:06:12
of the 80 20 rule and this idea of the
00:06:13
pareto chart and what he told us and
00:06:15
what he taught us is that
00:06:17
a pareto chart helps you separate the
00:06:19
vital few
00:06:20
from the trivial many now what did joran
00:06:22
mean what he means is when you're
00:06:24
solving a problem there's often
00:06:25
one or two key issues key root causes or
00:06:29
key
00:06:30
defects that you need to focus on to
00:06:31
have a major impact
00:06:33
on that particular situation and that's
00:06:35
exactly what you see here
00:06:36
when we take our data from the check
00:06:38
sheet and we put it into this pareto
00:06:39
chart
00:06:40
we see that control pcb issues accounts
00:06:43
for
00:06:44
nearly 40 of our defects you can see if
00:06:46
we come across here
00:06:48
we've got 40 percent of our defects
00:06:49
coming directly from control pcb
00:06:52
now there's two things happening on this
00:06:53
graph obviously there's the blue bars
00:06:54
which are simply just the frequency or
00:06:56
the count
00:06:57
of defects that occurred throughout the
00:06:59
week and then this black bar is actually
00:07:01
the cumulative
00:07:02
line so this first defect accounts for
00:07:04
40 percent and then we go up and up and
00:07:05
up all the way to 100
00:07:07
now that we have this pareto analysis we
00:07:09
know that control pcb
00:07:11
is our primary issue it tells us what to
00:07:13
focus on now we still don't understand
00:07:15
why
00:07:16
these issues are happening and this is
00:07:17
where something like the cause and
00:07:19
effect diagram can be incredibly useful
00:07:21
so this is the the fish bone diagram or
00:07:23
the ishikawa diagram
00:07:24
there's all sorts of different names for
00:07:26
it but it is a cause and effect diagram
00:07:28
and the way this works is we start with
00:07:29
the effect
00:07:30
that's over here on the right that's the
00:07:31
head of the fish here in orange this is
00:07:33
our effect
00:07:34
and so step one of the cause-and-effect
00:07:36
diagram is to start with a really
00:07:39
well-written problem statement
00:07:40
so i've put in pcb failures but in
00:07:42
reality you want to have a
00:07:44
much more descriptive problem statement
00:07:46
than this and once you have this effect
00:07:48
you can start working through the the
00:07:50
fish bone process
00:07:52
to analyze all of the potential causes
00:07:54
and failures
00:07:55
now i'm showing here what's called the
00:07:57
8ms and this is the beauty of the
00:07:59
fishbone process
00:08:00
is that it's a well-structured approach
00:08:02
to root cause analysis
00:08:04
it forces you to think about all of the
00:08:06
potential
00:08:08
different categories or scenarios or
00:08:10
causes that might be contributing to
00:08:12
your problem
00:08:13
now along with the cause and effect
00:08:15
diagram are a number of tools that you
00:08:16
should be using
00:08:17
so i would recommend you get out your
00:08:19
flow chart look at your process
00:08:21
use your flow chart and and ask yourself
00:08:23
how might each step in the process fail
00:08:26
and contribute to the the effect that
00:08:28
we're seeing teamwork is also a must
00:08:30
here
00:08:30
you're not going to be a subject matter
00:08:32
expert in all of those eight m's and you
00:08:34
need people from operations and
00:08:35
engineering and quality and r d
00:08:38
and marketing and maintenance to really
00:08:40
do a thorough analysis
00:08:42
in each of those areas to truly
00:08:43
understand the root cause and then of
00:08:45
course brainstorming
00:08:46
you know you're going to have to
00:08:47
creatively think about and talk about
00:08:49
and discuss
00:08:50
potential root causes that maybe you're
00:08:52
not even aware of
00:08:53
and then the five-way analysis i love
00:08:54
the five wise it really helps you go
00:08:56
from a high-level symptom
00:08:58
down to the true root cause and really
00:09:00
ask why why why
00:09:01
to truly get to those those real root
00:09:03
causes that you need to address
00:09:05
and then as you have that team
00:09:06
discussion and you you go through the
00:09:07
process
00:09:08
you can identify potential root causes
00:09:11
and contributing factors
00:09:12
to the problem you're trying to solve
00:09:14
now obviously again it's we have to go
00:09:16
back to that parade of principle we
00:09:17
can't focus on everything
00:09:18
we have to talk about the most likely
00:09:21
root causes and the most likely
00:09:23
contributing factors
00:09:24
so again at the end of your cause and
00:09:25
effect diagram you might identify three
00:09:27
or four issues that you need to study
00:09:29
further
00:09:29
now i wanna i wanna talk about this one
00:09:31
high humidity during assembly
00:09:33
now as we were working through the cause
00:09:34
and effect diagram process the engineer
00:09:37
who was helping us
00:09:38
looked at our check sheet and noticed an
00:09:40
interesting pattern
00:09:42
what they noticed here and i've
00:09:44
highlighted here in yellow is that
00:09:45
sunday monday tuesday we
00:09:47
we only had a few defects right six and
00:09:49
four and one whereas on wednesday
00:09:51
thursday friday you'll notice that our
00:09:53
defect rate jumped up a little bit
00:09:54
and what the engineer remembered is that
00:09:56
we had a rainstorm come through
00:09:58
on tuesday night and the humidity level
00:10:00
in the facility really jumped up
00:10:01
and so what the hypothesis here is that
00:10:04
humidity
00:10:05
is affecting our defect rate so i've
00:10:07
created this little table here to show
00:10:09
the days of the week
00:10:10
along with the defects and the humidity
00:10:13
now to truly understand this
00:10:15
relationship
00:10:15
we have to create a scatter diagram
00:10:19
so here's exactly what that scatter
00:10:20
diagram looks like
00:10:22
what we do here is we're plotting pairs
00:10:24
of data so for example on sunday we had
00:10:26
six defects
00:10:27
and 18 humidity you can see that right
00:10:30
here that's this data point right here
00:10:31
we had six defects
00:10:32
18 humidity now the way this scatter
00:10:35
diagram works
00:10:36
or you might hear this called an xy
00:10:37
scatter plot is here on the x-axis
00:10:40
we put our controllable variable our
00:10:42
independent variable
00:10:43
and then on the y-axis we put our
00:10:45
response variable so here we believe
00:10:47
that
00:10:47
relative humidity is the the independent
00:10:50
variable that is affecting
00:10:52
our response variable which is defects
00:10:54
and you can see here that there appears
00:10:56
to be some relationship
00:10:58
between pcb failures and humidity
00:11:01
now it's really important when you're
00:11:03
looking at the scatter diagram not to
00:11:04
assume that this relationship
00:11:06
is a causal relationship right there's
00:11:08
this really important concept that you
00:11:09
can have
00:11:10
correlation without causation two
00:11:13
parameters or two variables can
00:11:14
correlate
00:11:15
without having a cause and effect
00:11:17
relationship so let's assume though
00:11:19
let's assume that we've done a doe here
00:11:20
and we've proven
00:11:22
that humidity has an effect on our pcb
00:11:25
defects
00:11:26
we could come back to the scatter
00:11:27
diagram we could say okay
00:11:29
our target for pcb defects is five or
00:11:32
less let's call it let's call it five or
00:11:34
less
00:11:34
and so we come down here to humidity and
00:11:36
say okay we wanna control
00:11:38
humidity to around 20
00:11:41
to keep our defects low does that make
00:11:43
sense and that's a this is a great way a
00:11:45
scatter diagram is a great way to
00:11:47
understand the relationship
00:11:48
between two possible variables now once
00:11:50
you've done your scattered diagram
00:11:52
you can quantify the relationship
00:11:54
between those two variables
00:11:55
so what i'm showing here is the pearson
00:11:57
correlation coefficient
00:11:58
and this coefficient ranges from
00:12:00
positive 1 all the way
00:12:02
over here on the left to negative 1 all
00:12:04
the way over here on the right
00:12:05
and that ranges from a perfectly
00:12:07
positive correlation here you can see
00:12:08
that as x changes y changes
00:12:10
identically and then same thing here
00:12:12
with r equals minus one this is a
00:12:14
perfect negative correlation
00:12:16
now as we get closer to zero we start to
00:12:18
lose that relationship
00:12:19
so an r value of zero means there's no
00:12:22
correlation between those two parameters
00:12:24
as x changes
00:12:26
y basically does whatever it wants
00:12:27
there's no relationship between those
00:12:29
two variables
00:12:30
now the next thing we could do in our
00:12:31
analysis is to look at
00:12:33
relative humidity over time so let's say
00:12:36
we go out
00:12:37
we talk to our facilities engineers we
00:12:38
say okay give us the relative humidity
00:12:40
within our environment
00:12:42
you know every six hours for the last
00:12:44
six months and we can take that data and
00:12:46
we want to plot it because we need to
00:12:48
understand
00:12:49
how relative humidity is changing within
00:12:51
our facility
00:12:52
and one of the ways you could analyze
00:12:54
that data is with a histogram
00:12:56
so a histogram is just a very simple bar
00:12:58
chart that graphs the frequency of
00:13:00
occurrence
00:13:01
of continuous data and again this is a
00:13:03
great way to talk about your process
00:13:05
every process or every product or every
00:13:07
quality attribute out there
00:13:09
has some level of random normal
00:13:11
variation
00:13:12
that will often occur in a pattern and
00:13:15
as engineers we need to understand what
00:13:17
is the pattern
00:13:18
associated with with our outputs or our
00:13:20
process and a histogram is a great way
00:13:23
to understand the pattern or the
00:13:25
variation in your process
00:13:27
now you might grab this data and you
00:13:28
might get like a skewed distribution or
00:13:31
maybe a bimodal distribution
00:13:33
or exponential distribution there's all
00:13:35
sorts of distributions you might get
00:13:37
but it's great to know how your process
00:13:39
is behaving
00:13:40
now the other beautiful part about a
00:13:42
histogram is you can take this data
00:13:44
and let's overlay some some
00:13:46
specification limits right
00:13:48
now what we have is the beginnings of
00:13:50
process capability
00:13:51
so the histogram is a fantastic tool to
00:13:54
quantify and understand how your process
00:13:56
behaves
00:13:57
and if you compare that against the
00:13:58
specification limits we can now start
00:14:00
talking about process capability
00:14:02
okay so we're on to the very last and
00:14:04
final qc tool
00:14:05
let's assume we now control for humidity
00:14:09
and we want to make sure that that
00:14:10
change has been effective over time
00:14:12
a control chart is the right tool or the
00:14:15
perfect tool to do that
00:14:16
so what is a control chart it is
00:14:18
essentially a tool that allows you to
00:14:20
confirm that your process
00:14:22
is in control now when i say in control
00:14:24
what i mean is
00:14:25
that you're only experiencing normal
00:14:27
variation when your process is
00:14:29
experiencing normal cause variation
00:14:31
your data should fall with within those
00:14:34
control limits
00:14:35
by the way if you're new to spc i have a
00:14:37
whole separate video on control charts
00:14:39
you can go check it out
00:14:40
i've got both the x bar on our chart as
00:14:41
well as attribute data
00:14:43
and a control chart is a fantastic tool
00:14:45
to use at the end of a project
00:14:46
to monitor and control your process and
00:14:48
make sure that your changes were
00:14:50
effective
00:14:50
and let's take a look at what this looks
00:14:52
like for our particular process
00:14:54
so here's our process right the first
00:14:56
week of data you can see we're really
00:14:58
all over the place
00:14:59
and our control limits are really wide
00:15:00
because we're not controlling for
00:15:02
humidity
00:15:02
and we've got all this data and you can
00:15:04
see on average we have about
00:15:06
eight defects per day right we're really
00:15:08
jumping around here and then let's say
00:15:10
on day nine we start controlling for
00:15:12
relative humidity
00:15:13
and we've got our our control chart
00:15:14
we're collecting data and you can see
00:15:16
that for the next
00:15:17
you know 20 plus days our defect rate
00:15:20
has dramatically fallen in fact our new
00:15:23
mean defects per day is around three
00:15:26
so essentially we've gone from eight
00:15:28
defects per day down to three defects
00:15:30
per day
00:15:31
and we've hit our target of reducing
00:15:33
defects by 25 percent
00:15:35
we've gone from 20 plus defects a day
00:15:38
down to about 15
00:15:39
all by controlling relative humidity in
00:15:41
our process all right that's it for
00:15:43
today
00:15:44
i hope you enjoyed it if you did hit
00:15:45
that like button also if you're serious
00:15:47
about becoming a cqe
00:15:49
i've got a free course go check it out
00:15:50
it's at cqe academy.com
00:15:52
free course where i cover the top 10
00:15:54
topics on the cq exam
00:15:55
and i also give you a bunch of great
00:15:57
free practice exams to help you on that
00:15:59
journey
00:16:00
all right i hope you enjoyed it thanks
00:16:01
so much i'll see you again bye
