0% found this document useful (0 votes)
3 views11 pages

KNN and Decision Tree Model Insights

Uploaded by

ro6162
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
3 views11 pages

KNN and Decision Tree Model Insights

Uploaded by

ro6162
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Q For a KNN classifies as we increase

k does the model move


the value of
towards Quiz
over fitting
under fitting
Automated t Live Easy
Sol check dirt from
O K nearest neighbours
K defined by DS
hyper parameter

q
o
un s

Inti Eden
As we increase K what happens

Let K 100M
will always take majority class
under fit
d K 3 class green
o
o o
class Glee
K I

0
É
O O
o k t DB

o d
o
g O o
o

K Y
K 1
O
o o o
Q Your colleague has trained a Tree based
model to predict the marks
regressor
of students for a
very hard test
The Max marks for all exams are 100
Copen ended
Live Medium

Physics Maths Mechanics target


6 5 68 74 The model MADE
is 1 Can we

72 67 70 this into
put
35 40 43 production

Distributions
Mechanics
Physics

f
171717 maths
20 75
Let's understand how a DT work

TyAlldatapts

Do Do

To D Tsa
mean
var

FI there are other features aswell


I output mean more relaible
Fg it
2
mean
var variance the
B Do IDs Is
I Ia 43 In
what will be the old range
Q
U U2 Us Un

Yes there will be only h unique


values in regressor of in this case
Had realized this
you guys
so we
For regressions we use
RFR
bot finite unique values
get many
because it trees
averages many
never of
Tree based Regressor can

training data
anything outside of

target range Mechanics

Target has a range


75
d
of only 20

training data Linear regressor


in better
would be
Q How do you create ensemble model from
KUN and DT Copen ended

Ensemble
8164 35 3
a 260 IN Yohe
input accurate
Tess
accurate
fi KNN
ft DT

ce KUN is purely a classification


model we need to create ensemble for
Js

2 blue 3 green
y j
output Greet h blue 3 green

output Blue
Voting TI
keep odd number
Better Ideas Use Probabilities
DT Prob
KN N Prob
3
Adding Up 16 30 8
15
blue 1215 4161 2
14 7
green 315 216 2 30 15

output Blue
Q How can
you use
Log Reg for
a multi class problem statement
Live Easy Quiz

One Rest framework


Sol us
D D
it MI 1
gBxIx Red
III x elif M2 1
Blue
o x
X
X 88 O
o
else
0
080 f orange
DD
Rea vs Rest
Make 2 Logress
blue vs Rest
Q What is the probability of finding
simple G DCO
value Q
the optimum using
cost looks like this
if the function
2D view [Link] It
4 view
same [Link] I
Automated I Medium t
s
Quiz O OR Oz
501 If you init on flat surface G D will
10 10 100 61

IT 2
2
47T prob stuck

You might also like