Implementation of DT in Python | Python Courses in Tamil | Skillfloor - video Dailymotion

Skillfloor

Master the implementation of Decision Trees (DT) in Python with this step-by-step Tamil tutorial by Skillfloor!   Our Website: Visit 🔗 http://www.skillfloor.com  Our Blogs: Visit 🔗 https://skillfloor.com/blog/  DEVELOPMENT TRAINING IN CHENNAI https://skillfloor.com/development-training-in-chennai  DEVELOPMENT TRAINING IN COIMBATORE https://skillfloor.com/development-training-in-coimbatore  Our Development Courses: Certified Python Developer Visit 🔗https://skillfloor.com/certified-python-developer Certified Data BASE Developer Visit 🔗https://skillfloor.com/certified-data-base-developer Certified Android App Developer Visit 🔗https://skillfloor.com/certified-android-app-developer Certified IOS App Developer Visit 🔗https://skillfloor.com/certified-ios-app-developer Certified Flutter Developer Visit 🔗https://skillfloor.com/certified-flutter-developer Certified Full Stack Developer Visit 🔗https://skillfloor.com/certified-full-stack-developer Certified Front End Developer Visit 🔗https://skillfloor.com/certified-front-end-developer  Our Classroom Locations: Bangalore - https://maps.app.goo.gl/ZKTSJNCKTihQqfgx6 Chennai - https://maps.app.goo.gl/36gvPAnwqVWWoWD47 Coimbatore - https://maps.app.goo.gl/BvEpAWtdbDUuTf1G6 Hyderabad - https://maps.app.goo.gl/NyPwrN35b3EoUDHCA Ahmedabad - https://maps.app.goo.gl/uSizg8qngBMyLhC76 Pune - https://maps.app.goo.gl/JbGVtDgNQA7hpJYj9  Our Additional Course: Analytics Course https://skillfloor.com/analytics-courses https://skillfloor.com/analytics-training-in-bangalore Artificial Intelligence Course https://skillfloor.com/artificial-intelligence-courses https://skillfloor.com/artificial-intelligence-training-in-bangalore Data Science Course https://skillfloor.com/data-science-courses https://skillfloor.com/data-science-course-in-bangalore Digital Marketing https://skillfloor.com/digital-marketing-courses https://skillfloor.com/digital-marketing-courses-in-bangalore Ethical Hacking https://skillfloor.com/ethical-hacking-courses https://skillfloor.com/cyber-security-training-in-bangalore  #decisiontree #python #pythontutorial #skillfloor #tamilcoding #pythonintamil #machinelearning #datascience #datasciencetamil #machinelearningtamil #techeducation #mlalgorithm #supervisedlearning #classification #codingintamil #pythoncourse #tamiltech #visualization #pythonskills #sklearn #skillfloorcourse

Transcript

00:00Hello everyone, in the video we are going to talk about the implementation of decision tree in python.

00:07So in practical, we will talk about decision tree in python.

00:14So when we are going to talk about decision tree, we will talk about libraries.

00:19So we will talk about heart data set.

00:21So in the heart data set, we will talk about whether the person is having heart disease or not.

00:25Based on age, gender, chest pain type, 0, 1, 2, 3.

00:300 is anginal, 1 is non-anginal.

00:33So there are different types of CP.

00:36Then, Tres BPs.

00:38So BP level, cholesterol, ECG levels.

00:42So based on this, we will decide the target.

00:46So thal is called thalassemia.

00:51So there are different types.

00:53So these are completely encoded values in heart data set.

00:57So we will use this data set.

00:59So we will do first basic EDA checks.

01:02So the shape.

01:03So there are actually 303 rows and 14 columns.

01:06Then, if you check information,

01:08there are 303 normal values.

01:11So there are missing values.

01:12So in basic EDA checks, univariate analysis.

01:17So in univariate analysis, his plot, chest pain type, we will compare.

01:23Then, similarly, we have relationship plot with respect to chest pain and target.

01:28So our target is 0 and 1.

01:31So in 0 and 1, we will plot the chest pain types.

01:36Then, correlation analyze.

01:38So correlation on the target column,

01:40the highest set, we will contribute to the highest set.

01:43Like 0.43, minus 0.43, and minus 0.43.

01:48So this is X, old peak, and chest pain.

01:51So in the three columns,

01:52we will contribute to higher.

01:54Then, followed by that,

01:55we have CA thalassemia.

01:58CA thalassemia and slope.

02:00Then, we will check missing values.

02:02Then, we will check missing values.

02:04We will use null.sum.

02:06We will check missing values.

02:08We will check information already.

02:10In the data set,

02:11we will use duplicate value in this data set.

02:13That means, 14 columns of value

02:15are similar to replicate.

02:17So that's the duplicates value.

02:19So in hard.duplicate.sum,

02:21we will represent one data.

02:23So when we remove the data,

02:25we will use hard.dropduplicates.

02:28In place equal true, true,

02:30we will check the original data set.

02:33It will be affected.

02:38In the original data set,

02:39there will be changes.

02:40If we have 303 rows,

02:42we will check the duplicate value.

02:44We will check the shape.

02:46We will check the shape.

02:48Next,

02:49we will check the duplicate value.

02:51We will check the duplicate value.

02:53Then, his plot,

02:54we will analyze.

02:55This is right skewed data.

02:57Most people,

02:58in the range,

02:59we will analyze.

03:01Like 150,

03:03350,

03:05350.

03:07Then,

03:08we will analyze.

03:09With respect to

03:10gender and age.

03:12Gender and age.

03:13And with respect to

03:14target column.

03:15So target column provide.

03:16Legend box also.

03:17So here,

03:18we have 0,

03:19female and 1 male.

03:23And 0,

03:25person not having heart disease.

03:270 not having heart disease.

03:281 person having heart disease.

03:29So we will split up.

03:31So compare two.

03:32We will compare.

03:33Person not having heart disease.

03:352 gender is higher.

03:37Compared with person having heart disease.

03:42Next,

03:43we will check the decision tree.

03:45We will check the decision tree.

03:47So x axis is the target column.

03:49So x axis is the target column.

03:50Last column is the target column.

03:51So that we remove x.

03:52So x is equal to heart.i log.

03:53So x axis is the target column.

03:54Last column is the last column.

03:55So that we remove x.

03:56So x is equal to heart.i log.

03:58So all rows are going to be left.

03:59And last column is the last column.

04:01So colon minus 1 is the last column.

04:04Then y is the last column is going to be left.

04:06So heart.i log of all rows,

04:09minus 1.

04:10So if we check the x.columns,

04:13one of the target column is the target column.

04:15Then,

04:16we will try to train test.

04:18So we will split the independent and dependent features.

04:20Next,

04:21train test split use.

04:23X train x test.

04:24Y train y test.

04:25Split it.

04:26So test size is 25%.

04:29So 75-25% split.

04:31We will check ytrain.value count.

04:34That is,

04:35person having hard decision,

04:36123 repair.

04:37Person not having hard decision,

04:39103 repair.

04:40That is,

04:42we will check the individual.

04:43Because we will use the balancing.

04:45That is,

04:46we will check the concept.

04:48First,

04:49decision tree classifier.

04:50We will normally run.

04:51So that is,

04:52from

04:53sqlearn.tree,

04:54import decision tree classifier.

04:56So this is tree based algorithm.

04:58We will import decision tree classifier.

05:00This is classification problem.

05:02This is regression deal.

05:04Decision tree regressor use.

05:06Then,

05:07metrics import.

05:09Because accuracy check.

05:10Then,

05:11model equal to

05:12decision tree classifier.

05:14Random set equal to

05:1542.

05:16Just to provide.

05:17Here,

05:18criterion says,

05:19one keyword.

05:20So criterion,

05:21we have already

05:22two criterion.

05:23One,

05:24Guinea.

05:25Second,

05:26entropy.

05:27So criterion equal to

05:28Guinea.

05:29That is,

05:30we will provide.

05:31So,

05:32we will change.

05:34Criterion.

05:35equal to

05:42Guinea.

05:43And then,

05:44followed by splitter.

05:45Splitter,

05:46we will provide.

05:47We will provide.

05:48Then,

05:49followed by splitter.

05:50Splitter,

05:51we will provide.

05:56So,

05:57we will run.

06:00Okay.

06:01So,

06:02we have the same accuracy.

06:03The most are.

06:04So,

06:05we will do this.

06:07Okay.

06:08So,

06:09Decision tree of criterion.

06:10Guinea based.

06:11Splitting on the best.

06:12Random state.

06:13We have 42.

06:14Okay.

06:15Then,

06:16we provide training data.

06:17That is,

06:18x-train and y-train.

06:19Then,

06:20model predict.

06:21x-test use.

06:22That is,

06:23y-prediction.

06:24Then,

06:25accuracy score,

06:26we will compare.

06:27y-test to y-prediction.

06:28If we compare.

06:2980.2%

06:30we will get.

06:31Okay.

06:32Next,

06:33we will do heatmap analysis.

06:35So,

06:36heatmap analysis.

06:37y-test

06:38and y-prediction.

06:39So,

06:40y-test and y-prediction.

06:41Confusion matrix.

06:42we will compare.

06:4361.

06:44correctly classified data.

06:45So,

06:46diagonal elements.

06:47correctly classified data.

06:48So,

06:4915.

06:50misclassified data.

06:51So,

06:52precision,

06:53recall,

06:54f-1 scores.

06:55next.

06:56Next,

06:57we will compare.

06:58Smote analysis.

06:59So,

07:00smote analysis.

07:01Synthetic.

07:03stands for synthetic.

07:09Minority.

07:13Oversampling technique.

07:15basic.

07:16What we need to do is

07:19target column.

07:20For example,

07:21target column.

07:22Person not having heart disease.

07:24Having heart disease.

07:25Consider.

07:26So,

07:27person not having heart disease.

07:28200.

07:29Person having heart disease.

07:31100.

07:32That's the same.

07:33As you can see.

07:34So,

07:35we can see.

07:36In other words,

07:38we can see.

07:39duplicate data.

07:40That's the same.

07:41duplicate data.

07:42Supporting.

07:43In the person having heart disease.

07:44To generate.

07:45If we check in 0, we will check in 200.

07:49If we check in 100, we will be able to analyze synthetic data.

07:54If we have a person having heart disease, we will be able to generate almost 190 data.

08:02If we have 0 person not having heart disease, we will generate.

08:08We will create a balanced learning.

08:12First, we import from imbelearn.oversampling, import smote.

08:19Then smote import from sklearn.tree, decision tree classifier import.

08:26Then smote create.

08:29First smote equal to, smote of random state equal to 42.

08:33Then we create a model.

08:36Xtrain resample and ytrain resample.

08:39First, we will analyze the training phase.

08:44Because in the training phase, we will generate synthetic data and learn.

08:48What do we do?

08:49Xtrain and ytrain directly parse.

08:52Xtrain resample and ytrain resample create.

08:56We will create two keywords.

08:58What do we do?

08:59What do we do?

09:00Smart.fit.resample, xtrain, ytrain.

09:04So, when we apply smote apply, we will balance everything.

09:08This is the actual data.

09:10The actual data means that 123 people are not having heart disease.

09:15103 people are having heart disease.

09:17So, we will balance this.

09:20So, 0, 123.

09:211 is almost 123.

09:221 is almost 123.

09:23We will generate 130 samples.

09:26So, almost closer samples.

09:28We will generate training phase using smart.

09:32Then, if we use smart,

09:35we will fit the model.

09:37Then, xtrain resample and ytrain resample.

09:40So, first model equal to decision tree classifier create.

09:44You can classify it equal to balance.

09:46So, we will balance classification.

09:48Then, random weight equal to 42 provide.

09:51Then, model.fit.off.

09:53What do we use fit?

09:54That is xtrain and ytrain resample values.

09:58Smart analyze the variables.

10:01Then, xtrain predict accuracy check.

10:05First time, 80.2.

10:08Small analysis is 82.8.

10:11So, 1.3% rise.

10:13Okay.

10:14So, we will use small technique.

10:17Next, we will check confusion matrix.

10:21Next, we will check hyperparameter tuning.

10:26Hyperparameter tuning.

10:27So, hyperparameter tuning is decision tree classifier.

10:30What do we pass?

10:32Hyperparameter tuning is basically best parameters.

10:36Randomized search cv.

10:38So, grid search cv.

10:39Basically, we have different parameters.

10:40So, grid search cv.

10:41So, grid search cv.

10:42Basically, we have different parameters.

10:44We have different Cartesian products.

10:46We have different possibilities.

10:48Check the best parameters.

10:49So, first.

10:50Decision tree classifier model.

10:51So, first.

10:52So, first.

10:53So, first.

10:54Decision tree classifier model.

10:55So, first.

10:56Create Cv.

10:57So, first.

10:59Then, we will pass the parameters.

11:00So, that's the decision tree classifier model.

11:02Okay.

11:03So, the first argument.

11:04Rig search cv of tree underscore clf.

11:07That's the decision tree model.

11:08Okay.

11:09That is the first argument.

11:10Then, params.

11:11Params.

11:12Params on the path.

11:13Params. Params, we create decision tree.

11:16So tree create, we check what criteria we have,

11:19what parameters we have,

11:21like criteria, spitting criteria we have,

11:23we need to enter of here.

11:25Then spitting, best and random.

11:28Maximum tip, we provide lists,

11:301-10 pull.

11:32Then minimum sample spit.

11:34Sample spit is now,

11:36we spread samples.

11:38If we control this,

11:39we provide minimum sample spit.

11:41Then final leaf node,

11:43we control minimum samples leaf.

11:45Okay.

11:46So five parameters,

11:48we pass.

11:49This is the second argument.

11:52Then third argument is,

11:53classification problem deal.

11:55So scoring equal to accuracy,

11:57we provide accuracy.

11:58So accuracy paste,

11:59we provide best parameters.

12:01Then end jobs equal to minus,

12:03this basic parameters,

12:05so parameters first,

12:09we have to consider.

12:10Guinea, best.

12:11Okay.

12:12Then,

12:13Guinea, random.

12:15Then entropy,

12:19best.

12:21Entrophy, random.

12:24So different possible combinations.

12:26So different possible combinations.

12:28Now,

12:29first two categories,

12:30we provide different possible categories.

12:32Next three categories,

12:34Guinea,

12:35best.

12:36Depth,

12:37we provide from 1 till 10.

12:39So depth 1,

12:40then 2,

12:413,

12:42till 10.

12:43Then,

12:44Guinea,

12:45randomly,

12:461,

12:472,

12:48till 10.

12:49So different possibilities,

12:50values,

12:51consider.

12:52parameters,

12:53choose.

12:54Decision tree,

12:55create.

12:56Decision tree,

12:57create.

12:58Accuracy based,

13:00result.

13:01High or low,

13:02or low,

13:03we choose.

13:05Next,

13:06end jobs equal to minus,

13:08different combinations,

13:09try.

13:10Hardware support,

13:12so end jobs equal to minus,

13:14provide.

13:155,

13:16provide.

13:17So,

13:18actually,

13:19lines,

13:20provide.

13:21O was.

13:22Then,

13:23cv equal to 5,

13:24cross validation.

13:25Cross validation,

13:26entire data set,

13:275,

13:28divide.

13:29So,

13:30first time,

13:31first block,

13:32testing phase,

13:33training phase.

13:35Then,

13:36second time,

13:37run now,

13:38training phase,

13:39testing phase,

13:40training phase,

13:41training phase.

13:42in the possible combinations,

13:50with respect to accuracy.

13:52Then,

13:53x-train and y-train

13:54pass,

13:55best parameters,

13:56extract.

13:57Best parameters,

13:58criterion entropy,

13:59maximum depth,

14:00minimum samples,

14:018.

14:02So,

14:03decision tree,

14:04classify,

14:05in the dictionary,

14:06point,

14:07double pointer method,

14:08use the best parameters,

14:09extract.

14:10Then,

14:11training data,

14:12testing data,

14:13and accuracy.

14:14So,

14:15accuracy check,

14:1682.89.

14:17Then,

14:18we have different best parameters,

14:22so,

14:23actually,

14:24highlighted values,

14:2688%

14:27so,

14:28we will comment,

14:30next time,

14:31run now,

14:32directly,

14:33grid sets,

14:34run now,

14:35these parameters,

14:36run now.

14:37So,

14:38here,

14:39best parameters,

14:40acquire,

14:41here,

14:43complete decision tree,

14:45so,

14:46complete decision tree,

14:47hyper parameter tuning,

14:48more technique,

14:49here,

14:50implementation of decision tree,

14:51in python,

14:52next video,

14:53thank you.

Implementation of DT in Python | Python Courses in Tamil | Skillfloor

Category

Transcript

Recommended