Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
56 views1 page

ML Intro

1. The machine learning workflow involves data gathering, pre-processing, model selection, training and testing, and evaluation. 2. Machine learning is used when large volumes of data require computational analysis or when human insights are inadequate to infer patterns. 3. Machine learning differs from regular computing in that it aims to discover patterns and logic from past data rather than implementing predefined logic.

Uploaded by

smartanand2009
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
56 views1 page

ML Intro

1. The machine learning workflow involves data gathering, pre-processing, model selection, training and testing, and evaluation. 2. Machine learning is used when large volumes of data require computational analysis or when human insights are inadequate to infer patterns. 3. Machine learning differs from regular computing in that it aims to discover patterns and logic from past data rather than implementing predefined logic.

Uploaded by

smartanand2009
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
You are on page 1/ 1

Machine Learning Work Flow Learning: Systems that improve performance

Data Gathering through experience. Machine learning is about


Normalization figuring out and optimising the relationships
Numeric Standardization between pairs of variables and using those
Pre-Processing Imputation relationships or rules to predict future behaviour.
Label Encoding Experience: Historical data
Categorical
Dummification What is learnt: Rules, map inputs to certain
Supervised
Model Selection outputs
Unsupervised
Training &
How: From experience and data, optimize rules
Testing and relationships continuously
Evaluation

When to use ML : Computational (Large volumes) & Inferential (human insights are inadequate)

ML vs Regular computing:
 Implement a given logic Vs figure out the logic based on past data and trends
 Rules are already known Vs Rules are discovered
Type of ML

SUPERVISED LEARNING Supervised learning is used when we have knowledge of the expected
output and the features that contribute to that output i.e. the output labels are known.
 learning guided by human observations and feedback with known outcomes
 Labelled data provided in the form of past input and output pairs to teach the model
 Mapping rules between a set of inputs and output/s are learnt
 Needs reliable and unbiased data samples during training to be effective
For Continious data output, a Regression model is used
For Categorical data output, a Classification model is used
Un-Supervised Learning
 no feedback or right outcomes known to the machine in advance (No pre-defined labels)
 enables to discover patterns within data that are previously not known E.g. Clustering & PCA
 Clustering output can be further analysed after intitial groups are discovered

Targeted marketing, Recommended Systems, and Customer


Unsupervised Learning: Clustering
Segmentation are applications in
Fraud Detection, Image Classification, Diagnostic, and
Supervised Learning: Classification
Customer Retention are applications in
Real-Time decisions, Game AI, Learning Tasks, Skill Acquisition,
Reinforcement Learning
and Robot Navigation are applications in ...
Predicting one variable on the basis of
The main purpose(s) of (LR) is/are another
Explaining one variable in terms of another
Limitation of all regression techniques is that it can predict but
TRUE
never be sure about underlying causal mechanism.

You might also like