How to make machine learning models with decision trees?

What functions which we directly call They give the ability that citizen input to that function and use us you know So I’ve been trainee wise towards the end off of do also through the end of our certification how you can create your own functions and call every day You need it What if we do it now Then every time you will use this only rather than Cody So we do not encourage that has all such okay so in December convenient to use from uses from Decision Tree to bagging all the algorithms.


If you have a gut feeling that years around importers will work the best if we have that kind of in-depth edge of the data and death knowledge of random Forrester In that case I will say yes to the C I For example if you look at my data set when he does it has so many 123 foods fights 789 10 11 At what 11 ways I can form my decision Tree So my entropy calculation information in calculation Birnam Forest randomly are pictures So definitely you can understand that these things are meant for Decision Tree and themthest one of the writing off data here Toby choosing between but in the same case.


Let us see if you have got only this much of big This was your total If we ignore this part of it in this case you don’t have muchmanyions So decision 304 hours will be almost similar Agreed So in this case bagging and boosting will a judge It will be people Got it So this is the way you can You make your own analysis and try to do it But I would always scientist It does not take much time Toe check so better for checking purposes Implement all that God stuff even on the list So help Yes One question this.


These courts are almost period and they said can call to the function What is the actual rule of the beat analyst in that industry What does he have created was to do You know this is almost automatic nature So I’ll actually be exactly Adidas signed So they die Unless you do see the first thing which we have not like We have not discussed the so far because we are still learning face So the first thing what we do is validate over you people know how we validate their data Has anybody tried doing that.


If we talk about if you remember my first section where we had Sunday balls are with us in that case you remember he Wthes continuously asking you give why it should be normal Onda Why there should not be any correlations are my equal inanity should not So there are so many attentions that but we usually don’t use it So a data analyst or a data scientist basic level is is a dinosaur Your data scientists will basically take this kind of approach where I will see that whatever date I have visit fit will be put inside machine Young first check to be duck Okay so a second check will be your data cleaning So let us say you make modifications in your data again You have to go invalidate.


What I usually do not always know Date answered And this is my change Because that’s how similar this every change the shape of the data between the meeting off the data have which in the standard division of dictator I between the mean of the data have I changed the i Q R’s off the razor Lord off Been going check to validate If there is a huge amount of the friends between both of them then I will say whatever I’ve done here now it is garbage Please got it Yeah Yeah Next part of the data scientist is basically toe Um what I say is uh paper decision on which independent variables are to be used which I will be ignored because we’re still not touchdowns Overweight Lunch right That’s why you wears a little country Islam how I can remove it So that’s the reason you are is a little little stuck over that A war is the difference between distant So once I complete your once we complete complete ml the whole ML I do Then we’ll enter feature rises And in that feature is it And I will show you guys this matter not to take anything blind He is right.


Now we’re in a learning cough where they are not doing any revelation investing But once we learned I’ll also show you how to value get and test your stuff Okay Next next five comes in uh stress testing So do we know performance testing stress testing every heard this terms you face the software partly used Yeah experiencing off the way we will check the maximum limited mandate until it breaks Now Something else So in our model also the EMR ml model that will put we’ll do this kind of tests on that stress tests And you will come to know that if I yes this morning onto this data what is the lowest accuracy that I really get And what is the highest accuracy that I can You This is again a part of utilization morning, Okay Next I go and do validation on medic.


Let us say I completed my machine learning Let us say a completed all of these techniques and the best Just now we found out best regard was bagging with those 89.8 something for example Is it possible to improve this accuracy in future They see I got 89.8 on my training Did But tomorrow if my data changes let a sick line will use its own data Every temperate will not use the training that always uses their own testing date Our futuristic Is it possible to increase this There was a possible So by 1% we have to do some calculations So this is very data Scientists will come to picture the scaling the do it getting option There used Yeah they do skating Definitely.


When we provide them that does this model is tuned on the scaling part Definitive But that that does not mean that is going to improvise accuracy Well yes I hope you want my point C There are so many patterns inside the deal So let us say in claiming they doesdoIf you remember 567 a three and four If you remember it was around 40 Do boys and 30 dopers Not tomorrow This might not be a factor within the climb data on a vigilance dictator It right might not be division like this So definitely when there is no by a singer doll there are chances that back to this is going to go I nothing wrong with that But the client will always ask you by how much.


In that case of leaders and just will exactly know how to get this thing done Using statistics probably the status takes and will that really use will be Our concern is that they’re doing interrogations Yes exactly Exactly So a data analyst what he will lose I will tell you a very very clear cut difference when our data analyst Our data in junior and a data scientist So a data scientist is basically a universal I will call it a universal sect He knows completely what these two people our data analyst is in our if you take the Julies off our current market and in time this is something like the reporting specialist or a taboo specialist or a person was working on Kulik reporting all.


If you heard about these truths or a person was working on purely on to be a business intelligence where his job is to just find out some unfortunate is from the data and represented in the farm off graphs and numbers Okay he might be using this technique nothing wrong we did But these techniques he will already have it He’s not going to make muchmanynges on that because from the statistical point off you he’s not that strong Okay next is a data ingenious.

Leave a Comment