00:00In this video, we will take an introduction to decision tree.
00:05So, decision tree is a supervised machine learning algorithm.
00:10So, we will have this target column present at all.
00:13So, in the decision tree, both regression and classification tasks are handled.
00:19So, if we a regression task deal with decision tree regressor,
00:24now, we will have this classification task,
00:27decision tree classifier.
00:30So, in case of linear and logistic algorithm
00:33in case of linear and logistic algorithm,
00:34we can use a best fit line or sigmoid curve.
00:36So, we can use an equation based algorithm
00:39to model predictions.
00:40But in decision tree,
00:42we can follow tree based structures
00:45and we can predict final probability
00:47and possible outcomes.
00:50So, in one part,
00:52we can say node.
00:54So, in one part,
00:56we can say decisions.
00:58So, if we say decision notes,
01:01we can say overall notes.
01:03So, overall notes based on decisions based
01:05we have final model predictions.
01:07That is whether that particular person
01:09loan provides or not.
01:11So, yes or no.
01:12So, if we say decisions,
01:13we can follow tree based structure
01:15we can decide.
01:18Next is key features.
01:20So, key features of decision tree.
01:22First is hierarchy of decisions.
01:24So, one decision tree
01:26tree one and then,
01:27one entire data set
01:28as smaller subsets
01:29we can predict.
01:30Through sequence of decision notes.
01:33So, in the decision notes
01:34in the decision notes,
01:35we can further split
01:36and split and then
01:37we can easily interpret
01:38the final decisions.
01:40For example,
01:41if one the house price
01:42predict
01:43first is based
01:44based on number of bedrooms.
01:46So,
01:473,
01:484,
01:497,
01:508.
01:51So,
01:52different possibilities
01:53from bedrooms 3
01:54we can check
01:55we can see
01:56next
01:57we can see
01:59yes or no.
02:00Then,
02:01here are the features.
02:03So,
02:04one of the features
02:05we can consider
02:06and split
02:07this is the hierarchy of decisions.
02:10So,
02:11we can create a tree structure.
02:13Finally,
02:14the house price
02:15is $53.85
02:18in terms of
02:191000
02:20we can sell
02:21we can see
02:23next is
02:24recursive partitioning.
02:25So,
02:26recursive partitioning
02:27we can split
02:28the bedrooms
02:293,
02:304,
02:317,
02:327,
02:338.
02:34So,
02:35bedroom features
02:36analysis
02:37we can check
02:38we can see
02:3934,
02:407,
02:418.
02:42So,
02:43each and every
02:44parts
02:45are
02:46separation
02:47of data
02:48that
02:49is
02:50different
02:51and
02:52we can see
02:53different
02:54categorical
02:55data
02:56or
02:57numerical
02:58data
02:59we can split
03:00so,
03:01overlay
03:02features
03:03values
03:04and
03:05split
03:06then
03:07we can see
03:09structure
03:10data
03:11or
03:12a
03:13normal
03:14distribution
03:15so,
03:16we can see
03:17specific
03:18criteria
03:19and
03:20we can see
03:21non-linear
03:22data
03:23easily
03:24handle
03:25high
03:26we can
03:27maintain
03:28next
03:31next
03:32decision tree
03:33different
03:34terminology
03:35first is
03:36root node
03:37root node
03:38is the base node
03:39so,
03:41so,
03:42here
03:43we can see
03:44internal nodes
03:45and say
03:46one node
03:47or another node
03:48we can connect
03:49parts
03:50branch
03:51so,
03:53this level
03:54is
03:55depth
03:56equal to
03:571
03:58then
03:59this
04:00is
04:01depth
04:02equal to
04:032
04:04depth
04:051
04:06depth
04:07equal to
04:082
04:09next
04:10first
04:11consider
04:12this
04:13parent node
04:14and
04:15child node
04:16so,
04:17any node
04:18split
04:19parent node
04:20and
04:21child node
04:22consider
04:23this
04:24parent
04:25node
04:26and
04:27child
04:28node
04:29then
04:30tree
04:31based
04:32algorithm
04:33final node
04:34say
04:35leaf node
04:36say
04:37this
04:38spitting
04:39leaf node
04:40say
04:41next
04:42this
04:43split
04:44based
04:45on
04:46feature
04:47values
04:48young
04:49middle
04:50senior
04:51three
04:52different
04:53categories
04:54split
04:55so,
04:56split
04:57split
04:58criteria
04:59so,
05:00splitting
05:01criteria
05:02problem
05:03deal
05:04in
05:05infinity
05:06entropy
05:07so,
05:08spitting
05:09proper
05:10final
05:11decision
05:12next
05:13this
05:14regression
05:15task
05:16deal
05:17mean
05:18square
05:19error
05:20structure
05:21proper
05:22so,
05:23decide
05:24step
05:25all
05:26to
05:27decide
05:28to
05:29decide
05:30and
05:31make
05:33decide
05:35make
05:36more
05:37decisions
05:38and
05:39decide
05:40to
05:41decide
05:42how
05:43to
05:45decide
05:46to
