Things to know about machine learning.
                     Things to know about 
Machine Learning
What is ML?
• Machine Learning means a machine teaching itself on data from the previously collected Data 
by multiple analysis and regression.
• These Data can be structured, semi-structured or even semi-structured.
Fact 1
• Machine refers Your machine/computer and Learning refers to finding patterns from data
• Example:
1. Virtual Personal Assistants. ...
2. Virtual Personal Assistants. ...
3. Predictions while Commuting. ...
4. Videos Surveillance
5. Etc….
Fact 2
• Machine Learning is just Data + Algorithms, also known as the patterns of reading data, but 
Data is more important in this case.
• The data needs to be converted to information before processing.
Fact 3
• Extraction of the functionality is the key in ML. 
• If total predictive ability is 100% then feature engineering effort = 80% and the learni
ng algorithm effort = 20%.
Source: Google Images
Fact 4
• Over-fitting implies a model that too well models 
training data. 
• Over-fitting happens when a model absorbs the 
information and sounds in the training data to the 
detriment of the model's output on new data.
• Basically, in over-fitting, machine memorizes instead of 
learning.
Source: Google Images
Fact 5
• On the off chance that you have limited quantities of information, at that point you're in an 
ideal situation utilizing simple models (linear regression). 
• In the event that you have a lot of information you can evaluate progressively complex 
models (Deep Learning, and so on.)
Fact 6
• To abstain from overfitting, consistently use regularization. Utilizing the procedure of 
regularization, ML specialists attempt to diminish the unpredictability of the relapse work 
without really lessening the level of the basic polynomial capacity.
Source: Google Images
Fact 7
• Preparing is the most significant piece of Machine Learning. Pick your highlights and hyper-
parameters cautiously. 
Essentials: 
• Statistics.
• Linear Algebra.
• Calculus.
• Probability.
• Programming Languages like R, Python.
Fact 8
• Machines don't take choices, individuals do. 
• In AI settling on a choice can be seen as doling out or 
foreseeing right marks (for instance purchase, not to 
purchase) in light of information for the thing 
highlights. 
• Utilizing preparing information like above specialists 
train a classifier and afterward use it to pick the class 
(settle on choice) for new information.
Fact 9
• Data cleaning is the most significant piece of 
Machine Learning. 
• You know the adage: Garbage in Garbage out. 
• Today information researchers regularly wind up 
investing 60% of their energy cleaning and binding 
together messy information before they can apply 
any examination or AI.
THANK YOU