Machine Learning
Machine Learning is subset of AI ( Artificial intelligence)
Machine Learning Workflow:
Use Algorithms which figures out rules for us instead of you (developer) writing rules for each input.
INPUT -> Classifier ->OUTPUT
Classifier - we train classifier. Classifier is function which take data as input and assigns labels to it as an output.
e.g. decision is a type of classifier
"if classifier is box of rules than learning algorithm is procedure that creates them" since it finds patterns in training data
in scikit training algorithm is included as fit() function.
Supervised Learning:
collect & Train data -> Train Classifier -> Make prediction
Training Data:
feature 1, feature 2, .... feature n -> Label
e.g.
weight, Texture -> Label
150g, Smooth -> Apple
130g, Bumpy-> Orange
Ideal Features are :
The Pillars of Machine Learning :
optimization
data pre processing
types of learning
data-set splitting
model evaluation
Machine Learning Workflow:
- Understanding problem and objective
- Reading from data sources
- EDA - Exploratory Data Analysis
- Data Cleaning
- Modeling
- Deployment and Reporting
Use Algorithms which figures out rules for us instead of you (developer) writing rules for each input.
INPUT -> Classifier ->OUTPUT
Classifier - we train classifier. Classifier is function which take data as input and assigns labels to it as an output.
e.g. decision is a type of classifier
"if classifier is box of rules than learning algorithm is procedure that creates them" since it finds patterns in training data
in scikit training algorithm is included as fit() function.
Supervised Learning:
collect & Train data -> Train Classifier -> Make prediction
Training Data:
feature 1, feature 2, .... feature n -> Label
e.g.
weight, Texture -> Label
150g, Smooth -> Apple
130g, Bumpy-> Orange
Ideal Features are :
- Informative
- simple
- Independent
The Pillars of Machine Learning :
optimization
data pre processing
types of learning
data-set splitting
model evaluation
Comments
Post a Comment