This document provides a comprehensive introduction to machine learning (ML), covering definitions, types of ML (supervised, unsupervised, and reinforcement learning), and various applications and algorithms. It elaborates on essential concepts such as data types, handling missing data, and model performance metrics. Additionally, it highlights the relationship between features and prediction functions, emphasizing the importance of model evaluation techniques.

1. Machine learning intro
The only limit to AI is human imagination. - Chris Duffey
By : Anas Jamil
Mar - 2019

2. Agenda
1- AI & ML & DL.
2- Machine learning (ML) introduction.
3- Types of Machine learning.
4- ML data types.
5- Working with missing data
6- Model performance (fitting)

3. AI & ML & DL introduction.
Artificial Intelligence (AI) : It is the study of how to train the computers so that computers can do things
which at present human can do. -www.geeksforgeeks.org-
Machine learning (ML) :Is the scientific study of algorithms and statistical models that computer systems
use to perform a specific task without using explicit instructions. -wikipedia-
Deep learning (DL): is an artificial intelligence function that imitates the workings of the human brain in
processing data and creating patterns for use in decision making.

5. What is Machine Learning?
“Learning is any process by which a system improves performance from
experience.” - Herbert Simon
Some use cases: We ML when:
• Human expertise does not exist (navigating on Mars)
• Humans can’t explain their expertise (speech recognition)
• Models are based on huge amounts of data (genomics).

6. Machine learning (ML)

8. 1- Supervised (inductive/class driven) learning
Supervised learning: is the machine learning task of learning a function that maps an input to an output
based on example input-output pairs.
Supervised learning: is where you have input variables (x) and an output variable (Y) and you use an
algorithm to learn the mapping function from the input to the output.
Y = f(X)
The goal is to approximate the mapping function so well that when you have new input data (x) that you
can predict the output variables (Y) for that data.
It is called supervised learning because the process of an algorithm learning from the training dataset can
be thought of as a teacher supervising the learning process.

9. Supervised learning terms:
Class, Target, Label
Attribute, Feature
Labeled data, dataset,
Sample data
Sample
Example
Record
Row
Instance
observation

10. Supervised learning:

11. Supervised learning algorithm types:
1- Regression: It is a Supervised Learning task where output is having continuous value.(Numeric output )
Ex: how much home worth.
2- Classifications: It is a Supervised Learning task where output is having defined labels(discrete value).
A- Binary classification: Yes/No
Ex: Spam email?
B- Multi-classes: One out of several outputs.
Ex:What is the weather?
Sample of algorithms : Support Vector Machine (SVM), Random Forest, Linear Regression, Decision Trees

12. 2-Unsupervised (data driven) learning
Training data does not include desired outputs.
Unsupervised learning: is very much the opposite of supervised learning. It features no labels. Instead,
our algorithm would be fed a lot of data and given the tools to understand the properties of the data. From
there, it can learn to group, cluster, and/or organize the data in a way such that a human (or other
intelligent algorithm) can come in and make sense of the newly organized data.

13. Unsupervised learning
Unsupervised learning classified into two categories of algorithms:
Clustering: A clustering problem is where you want to discover the inherent groupings in the data,
such as grouping customers by purchasing behavior.
Association: An association rule learning problem is where you want to discover rules that describe
large portions of your data, such as people that buy X also tend to buy Y.

14. Unsupervised learning: Dimensionality Reduction (DR)
There are two components of dimensionality reduction:
Feature extraction: This reduces the data in a high dimensional space to a lower dimension space,
i.e. a space with lesser no. of dimensions.
a+b+c+d = e
ab+c+d = e
Feature selection: In this, we try to find a subset of the original set of features, to get a smaller subset
which can be used to model the problem
c = 0
Sample of algorithms : FastText alg, BlazingText alg, Principal component analysis (PCA)

15. 3- Reinforcement learning (RL)
It is about taking suitable action to maximize reward in a particular situation.

16. 3- Reinforcement learning (RL)
Types of Reinforcement: There are two types of Reinforcement:
1- Positive: Positive Reinforcement is defined as when an event, occurs due to a particular
behavior, increases the strength and the frequency of the behavior. In other words it has a positive
effect.
2-Negative: Negative Reinforcement is defined as strengthening of a behavior because a negative
condition is stopped or avoided.
Use Cases of RL : Real-time decisions , Game AI, Robo navigation, auto drive cars

17. Data types from ML perspectives
1- Numerical Data
2- Categorical Data
3- Time Series Data
4- Text

18. 1- Numerical Data:
Numerical data is any data where data points are exact numbers. Statisticians also might call numerical
data, quantitative data. This data has meaning as a measurement such as house prices.

19. 2- Categorical Data
Categorical data represents characteristics, such as a hockey player’s positions.
Categorical data can take numerical values. For example, maybe we would use 1 for colour red and 2 for
blue. But these numbers don’t have a mathematical meaning.

20. 3- Time Series Data
Time series data is a sequence of numbers collected at regular intervals over some period of time. It is
very important, especially in particular fields like finance. Time series data has a temporal value attached
to it, so this would be something like a date or a timestamp that you can look for trends in time.

21. 4- Text
Text data is basically just words. A lot of the time the first thing that you do with text is you turn it into
numbers using some interesting functions like the bag of words formulation.
We can use stemming, lowercase functions .. etc

22. 4- Text
This is working not disappointed
This is not working. disappointed
Tokenization :
[ ‘disappointed’, ’is’, ’not’, ’working’, ’this’ ]

23. 4- Text

24. 4- Text
Orthogonal sparse bigram (OSB) :

25. 5- Working with missing data
For row have a missing values you can :
1- Delete the row (if data is not related).
2- Impute missing data:
A- if data is related to each other you can calculate the mean for that column.
B- if data is independent you can pick data from another row.
C-if data is related to timestamp:
1- interpolation
2- fill backward
3- fill forward

26. 5- Working with missing data

27. 5- Working with missing data

28. 5- Working with missing data

29. 5- Working with missing data

30. 5- Working with missing data

31. 5- Working with missing data

32. 5- Working with missing data
Some useful lib for python:
1-Numpy : Mathematical function for optimize large data
2- Pandas : Data analyzing and modeling & reading
3- Matplotlib : Plotting library for visualize the data

33. 6- Model performance (fitting)
Relationship between input and output could be:
1- Liner
2- non-liner
Knowing this relation will help in using algorithm and choose the attributes needed in
predict function

34. 6- Model performance (fitting)
1- Underfitting:
When: Poor performance in testing set , poor in training set
Why: Feature is not enough to capture the relationship between input and output
How : Add more rows , or add more features, optimize the hyperparameters
2- Overfitting:
When: Poor performance in testing set , Good in training set
Why: Model memories the data it has seen and unable to generalize it on unseen data.
How: Removing complex feature and optimize the hyperparameters
3- Balanced:
Good performance in testing set , Poor in training set

35. Regression model performance
Common Techniques for evaluating performance:
Visually observe using Plots
Residual Histograms (negative less than positive)
Evaluate with Metrics like Root Mean Square Error (RMSE)

36. Binary & multi-class model performance
Common Techniques for evaluating performance:
Visually observe using Plots
Confusion Matrix

37. Binary model performance (SKLearn)

38. References:
https://docs.aws.amazon.com/machine-learning/index.html