Weka Tutorial 15 Java API 101 (Application) Video Lecture - Weka Tutorial

okay welcome to the 15th weapon tutorial
in this tutorial I'm going to show you
how you can interact with the Waker Java
API for the first time so in my last
tutorial I've shown you how to create a
project in Eclipse IDE for Java and how
to how to import the wicked jar file
that I was downloaded downloading from
the wicked download website so let's
start so in this tutorial you can see
that I already created a class called
start wake up and we have a main method
there so I'm going to I'm going to
illustrate the start wake a class step
by step every line I will describe this
and I hope that you will enjoy this
video and it will immediately start to
work with the waked Java API ok so so to
start with you can see that I have
imported numbers of packages like
bufferedreader file reader and random
and I think random is not used in this
kind in this tutorial so that's an
unused package here I also imported most
important importantly some of the
wake-up packages like classifiers and
especially evaluation and base with 9
base because we are trying to get some
results by applying an eye base
classifier I also included the wakeup
core package instances and then I
started writing the class start waking
and I'm having the main method here so
the bufferedreader a package was
imported because we have to you have to
read of an hour file from your drive
from from my drive ok so in order to do
that I just created a bufferedreader
object here so let me highlight that for
you with unread so the class is
bufferedreader and I have created the
object called be reader and initialize
that to no so what happens with this be
reader so be reader actually reads the
file on our file that you have created
and if you can recall from from the
first tutorial that it our file contains
the feature values that you extracted
from your data set by using some kind of
language or you can do it by hand or you
can even import it export it from an
excel sheet so now with this
bufferedreader here we are actually
reading this file
situated in my H Drive so for the sake
of this tutorial I'll go to H Drive and
outside - dot RM take a look what is
actually there so you can see that in my
art file I have lots of in lots of
features fog index fog index wsw and
with lots of values here so you don't
have to be bothered about these values
because we're just interested in in a
number of features here so we have
numbers numbers of features and in the
end you can see that we are having an
attribute called class yes or no so
that's my class attribute and this is
situated at the very last DC situated as
at the very end of my attribute section
or this is exactly the last attribute in
my feature set ok so now I'm going back
to my java code here that's important
break2 to illustrate together and to
explore the our file that we are
currently working with so now what we
have done here we have created an object
called train with the class instances
and we have sent a B reader
bufferedreader object there so what
we're trying to do here is actually we
are sending the to dot our file to this
to this particular object called train
and we are calling asset class index
method on the train dog Nam attributes
method - one what this line means is
that actually we are setting the class
attribute here so you can recall from my
to dot our file that this class
attribute was situated at the very end
of the attribute section in other words
this was the last attribute so what we
are doing here we have 40 attributes so
40 minus 1 so 0 to 39 are the numbers of
oops so 39 is going to be my class
attribute so what I'm doing here train
dot nom attributes this returns 40 minus
1 so that means the class index is going
to be 39 or in other words that's the
class attribute yes or no having the
values yes on oh so we have just said
the class attribute for my to dot our
file so now we have closed the be reader
because it's always a good practice in
Java to close your buffered reader
whenever you are finished or whenever
you've done with above reader by reading
some stuffs from your drivers or or what
else now we are creating an eye base
object here called NB of night base
class and this is this is the normal
this is the normal constructor we are
just calling here but here we are
calling a method called bit classifier
so what we are trying to do here we're
saying that ok this is the train set so
this is actually my train set and we are
sending that to NB for by calling the
method called build class affair now the
classifier will be built or induced on
on with this a to dot our farm ok so now
we are going down and ok this case the
screen ok so now we sent the train file
to NB by calling the method call build
classifier now we are again creating an
object of type evaluation the evolution
is a class that's defining the wakelock
api of like a jar and eval is my object
name and we are sending the parameter
and the train file the to dot our file
here is the parameter for creating that
particular object called eval now we are
we are calling the method called cross
validate model and here we are sending
numbers of parameters 1 2 3 4 in total
so the first parameter is NB that's the
night based classifier we have
created here and in this line and we are
also sending the second attribute train
that's our file we are saying okay we
are going to have 10-fold
cross-validation for creating that
particular model and we are also sending
a random seed called random seed with
the number integer number one so that
means that any you can start we are
saying that okay waiter you can start
with any one of the faults and that's
set to one okay so we are here about to
the end of the program so these are
these are the two lines that will be
shown on the output screen of eclipse
and system dot out dot println we are
calling your to summary string method of
eval that's simple that's just showing
you the results here and this will show
you the results of the error rates the
Kappa statistics or whatever the
statistics you were probably were so far
experiencing with the wake explorer so
those those results will be shown on the
output pin but not on waked eastern bar
on Eclipse results results pin output
pin and the second system dot out dot
println actually tail actually tells you
the if measure the position and the
recall that we found with with with
night base applied on this particular
our file called two dot ARF with a
10-fold cross-validation so that's
pretty much it pretty much of it the
code first go to interact with Java API
of wake us so now let's run this code so
we run it and let's take a look at the
result on the result pen I'm just
maximizing that result you can see the
currently classified instances 71% 28%
in currently classified instances we
have the kappa statistics and so much so
much of Weka Explorer that we're that
we're currently experienced by us in the
classify tab
root mean squared error relative
absolute error so if you can recall if
you really are with this tutorial you
can recall that these are the statistics
all the time shown on the Waker output
pin on the classify tab and these are
the values of F score point eight point
zero point eight two are point eight
three actually and this is precision
point nine seven and this is the recoil
point seven two so these are the if
score precision and recall of to dot our
file when applied in B or nine based on
it with the 10-fold cross-validation set
up so I think you'll be able to play
with wake-up Java API and you can change
everything you can change the setup
because it is set up was for 10-fold
cross-validation with nine bits of
course you can apply other classifiers
as well and you can change the setup
from 10-fold cross-validation to
training and test split 6040 or
something like that so this means that
with with this java api from where you
can do anything you could do with the
wake i explorer or wake a GUI so that's
pretty much a fit i think you enjoyed
this video and keep keep in touch thank
you