Scribd
Upload
1K views1 page

Great Learning Notes

This document discusses different categories of machine learning: supervised learning, unsupervised learning, and reinforcement learning. It provides details on supervised learning, including how it uses training and testing data sets to build models and predictive algorithms for classification and regression. Unsupervised learning is described as not having training and testing stages and instead finding hidden patterns in the form of clusters and associations within all input data.

Uploaded by

RohitSurya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
1K views1 page

Great Learning Notes

This document discusses different categories of machine learning: supervised learning, unsupervised learning, and reinforcement learning. It provides details on supervised learning, including how it uses training and testing data sets to build models and predictive algorithms for classification and regression. Unsupervised learning is described as not having training and testing stages and instead finding hidden patterns in the form of clusters and associations within all input data.

Uploaded by

RohitSurya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 1

Part of statistics that is used to understand the data is called Descriptive Statistics.

Part of statistics that is used to establish the reliability of potential patterns identified is
called Inferential Statistics.

Machine Learning categories:


 Supervised Learning
 Unsupervised Learning
 Reinforcement Learning

SUPERVISED LEARNING:
A data set with patient’s information such as age, BP level, sugar level, etc., are
called independent attributes (which we already have) and the ‘thing’ that we try to predict
is called target attribute (which we don’t have).
Algorithms that work in 2 stages – training and testing. The data set is split into 2 –
training set and testing set. For model building, only the training set is used.
To see if those models work in real life conditions or they’re just an artifact or a
statistical fluke, Alternate Hypothesis Testing is done. ‘Alternate’ to the Null Hypothesis. Null
Hypothesis aka Law of Internia which basically tells the model is wrong and there is no
relationship between the attributes.
Using inferential statistics we get statistical parameters such as p-values which tells
us the probability of the model created being a true model in the real world. Once the
inferential statistics supports the model, then the next thing to see is the accuracy of the
model in the real world.
Predicting numerical values – Regression Modelling
Predicting other things – Classification Modelling

UNSPERVISED LEARNING:
No training and testing stages. No independent attributes and dependent attributes.
All the values are fed into the algorithms and it tries finding the hidden patterns in
the form of clusters and associations reflecting some kind of commonality or togetherness.
It is the responsibility of the data scientist to analyze the clusters and give meaning to them.

You might also like