I. Introduction - Background

1 Why Machine Learning?
- 1.1 Problems Machine Learning Can Solve
- 1.2 Knowing Your Task and Knowing Your Data
2 Why Python?
3 scikit-learn
4 Essential Libraries and Tools
5 Further Reading

“In this chapter, we will explain why machine learning has become so popular and discuss what kinds of problems can be solved using machine learning. Then, we will show you how to build your first machine learning model, introducing important concepts along the way.” (pg.1)

1 Why Machine Learning?

“Machine Learning is about extracting knowledge from data.” (pg.1)

“Intelligent applications” that use expert-designed hardcoded rules have two major disadvantages:
- Requires a deep understanding of how the decisions are made by a human expert
- The system is not generalizable if the logic underlying the decision-making process is domain-specific

1.1 Problems Machine Learning Can Solve

“The most successful kinds of machine learning algorithms are those that automate decision-making processes by generalizing from known examples.” (pg.2)

	Supervised learning	Unsupervised learning
Description	The user provides the algorithm with pairs of inputs and desired outputs, and the algorithm finds a way to produce the desired output given an input.	Only the input data is known, with no known output data gven to the algorithm
Advantages	The algorithms are well-understood and performance is easy to measure	Detect previously unknown or uncertain patterns
Disadvantages	Creating a dataset of inputs and outputs can be	The algorithms are usually harder to understand and evaluate
Examples	- Identifying the zip code from handwritten digits on an envelope - Determining whether a tumor is benign based on a medical image - Detecting fraudulent activity in credit card transactions	- Identifying topics in a set of blog posts - Segmenting customers into groups with similar preferences - Detecting abnormal access patterns to a website

In both cases, the data need to be represented in a form understandable by the algorithm, namely in tabular form
- Each row, a “sample”, represents a data entity
- Each column, a “feature”, represents a property that describe these data entities
Feature extraction/engineering is a key part of building a good representation of the dataset
- As the ML algorithm cannot make predictions for which it has no information, such as predicting gender based on last names

1.2 Knowing Your Task and Knowing Your Data

One of the most important part of the ML process is understanding how the data relates to the problem at hand
- As each algorithm differs in terms of what type of data and problem it is best suited for
- Also keep in mind all the explicit and implicit assumptions that you might be making
When building a ML solution, keep the big picture in mind by asking these questions:
- What question(s) am I trying to answer? Do I think the data collected can answer that question?
- What is the best way to phrase my question(s) as a machine learning problem?
- Have I collected enough data to represent the problem I want to solve?
- What features of the data did I extract, and will these enable the right predictions?
- How will I measure success in my application?
- How will the machine learning solution interact with other parts of my research or business product?

2 Why Python?

Python has a wide range of libraries for data science
Can interact directly with the code using a terminal or the Jupyter Notebook
- This is important as ML and data analyses are iterative processes require easy interaction
Python can also be used to create graphical user interfaces and web services

3 `scikit-learn`

An open-source project that contains a number of state-of-the-art ML algorithms
pip is a quick and convenient option to install scikit-learn and its dependencies:

pip install numpy scipy matplotlib ipython scikit-learn pandas

4 Essential Libraries and Tools

4.1 Jupyter

An interactive environment for running code in many programming languages in the browser
A great tool for exploratory data analysis

4.2 NumPY

One of the fundamental packages of scientific computing in Python
The core functionality of NumPy is the ndarray class, a n-dimensional array of elements of the same type
scikit-learn takes in data in the form of NumPy arrays

import numpy as np
x = np.array([[1, 2, 3], [4, 5, 6]])
print(x)

## [[1 2 3]
##  [4 5 6]]

4.3 SciPy

A collection of functions for scientific computing in Python, including
- advanced linear algebra routines
- mathematical function optimization
- signal processing
- special mathematical functions
- statistical distributions

4.3.1 Create a 2D NumPy array with a diagonal of ones, and zeros everywhere else:

from scipy import sparse
  
eye = np.eye(4) 
  
print(eye)

## [[1. 0. 0. 0.]
##  [0. 1. 0. 0.]
##  [0. 0. 1. 0.]
##  [0. 0. 0. 1.]]

4.3.2 Convert the NumPy array to a SciPy sparse matrix in compressed sparse row (CSR) format:

sparse_matrix = sparse.csr_matrix(eye) 
  
print(sparse_matrix)

##   (0, 0) 1.0
##   (1, 1) 1.0
##   (2, 2) 1.0
##   (3, 3) 1.0

4.3.3 Create a sparse representation directly:

data = np.ones(4) 
  
row_indices = np.arange(4) 
  
col_indices = np.arange(4) 
  
eye_coo = sparse.coo_matrix((data, (row_indices, col_indices))) 
  
print(eye_coo)

##   (0, 0) 1.0
##   (1, 1) 1.0
##   (2, 2) 1.0
##   (3, 3) 1.0

4.4 `matplotlib`

The primary scientific plotting library in Python
Visualizing the data and different aspects of the analysis can provide important insights.

import matplotlib.pyplot as plt
  
plt.figure(1)
  
# Generate a sequence of numbers from -10 to 10 with 100 steps in between 
x = np.linspace(-10, 10, 100) 
  
# Create a second array using sine 
y = np.sin(x) 
  
# The plot function makes a line chart of one array against another 
plt.plot(x, y, marker="x")
  
plt.show(1)

4.5 `pandas`

A Python library for data wrangling and analysis
A pandas DataFrame is a table, similar to an Excel spreadsheet, on which a wide range of modifications and operations can be performed
Each column of the dataframe can be a different data type
Files/databases of many different types can be imported into a pandas dataframe, such as CSV, Excel, and SQL files

4.5.1 Creating a dataframe:

import pandas as pd 
import mglearn
  
# Create a simple dataset of people 
data = {'Name': ["John", "Anna", "Peter", "Linda"], 
  'Location' : ["New York", "Paris", "Berlin", "London"],         
  'Age' : [24, 13, 53, 33]        
  } 
  
# Convert data into a dataframe
data_pandas = pd.DataFrame(data) 
  
# Print the dataframe
print(data_pandas)

##     Name  Location  Age
## 0   John  New York   24
## 1   Anna     Paris   13
## 2  Peter    Berlin   53
## 3  Linda    London   33

4.5.2 Data in the dataframe can be easily selected/filtered:

print(data_pandas[data_pandas["Age"] > 30])

##     Name Location  Age
## 2  Peter   Berlin   53
## 3  Linda   London   33

5 Further Reading

Sparse matrices

Sparse Matrices in SciPy - SciPy Lecture Notes
A Gentle Introduction to Sparse Matrices for Machine Learning - Machine Learning Mastery by Jason Brownlee
Sparse Matrices For Efficient Machine Learning - Standard Deviations by David Ziganto

I. Introduction - Background

Part 1 of notes for Introduction to Machine Learning with Python by Muller and Guido

I. Introduction - Background

1 Why Machine Learning?

1.1 Problems Machine Learning Can Solve

1.2 Knowing Your Task and Knowing Your Data

2 Why Python?

3 scikit-learn

4 Essential Libraries and Tools

4.1 Jupyter

4.2 NumPY

4.3 SciPy

4.3.1 Create a 2D NumPy array with a diagonal of ones, and zeros everywhere else:

4.3.2 Convert the NumPy array to a SciPy sparse matrix in compressed sparse row (CSR) format:

4.3.3 Create a sparse representation directly:

4.4 matplotlib

4.5 pandas

4.5.1 Creating a dataframe:

4.5.2 Data in the dataframe can be easily selected/filtered:

5 Further Reading

3 `scikit-learn`

4.4 `matplotlib`

4.5 `pandas`