Machine learning applications

Machine learning:

• Supervised learning
• finite number of discrete categories --> classification problem
• continuous variables --> regression problem
• Unsupervised learning
• discover groups of similar examples --> clustering problem
• determine the distribution of data within the input space --> density estimation problem
• project the data form a high-dimensional space down to two or three dimensions --> visualization purpose

[CS61B-2015] Generics, Conversion, Promotion

An example of unboxing and autoboxing.

Import a module from your package in Python

If I have two modules in two packages in the same folder described below:
myfolder/
        package1/
                __init__.py  (need to add an empty file)            
                module1.py (has a function test())
        package2/
                __init__.py
                module2.py

I can use the following code in module2 to import module1

3 ways to prevent a class from being subclassed in Java

1. Access control: Even though a class can't be marked private, a class can be non-public (by not declaring it as public). A non-public class can be subclassed only by classes in the same package as the class.
2. final keyword: A final class means that it's the end of the inheritance line.
3. If a class has only private constructors. Ex. like built-in Math class in Java.

MIN_VALUE in Java

It is no difference between declaring a MIN_VALUE with a "-" notation or without a "-" notation in Java. Both of the variables will be -2147483648.

Bias vs Variance in machine learning

High bias -> underfitting

High variance -> overfitting

Reference: https://www.coursera.org/learn/machine-learning/home/welcome

Default initial value of variables in Java

For instance variables, Java has default initial value for those variables. But, the wrapper class of primitive variable, like Integer, Float..., will not have default initial value even they are declared as instance variables.

For variable within a method, Java has NO initial value for those variables.

Java Visualizer

"Java Visualizer" is a very helpful tool that can visualize the relationships between references and objects. See the following example.

Link: http://cscircles.cemc.uwaterloo.ca/java_visualize/

[CS61B-2015] Notes for class 5-Binary and Bits

8 primitive types of Java:

• The 5 integer types
• Exact representations of integers.
• byte, short, int, long, char.
• The boolean type
• The 2 floating point types
• Approximate representation of reals.
• float, double.

Integer:

bit operations:

[CS61B-2015] Notes for class 4-Testing

https://berkeley-cs61b.github.io/public_html/materials/lab/lab2/lab2.html

import org.junit.Test;
import static org.junit.Assert;
assertEquals(expected, actual);

To make testing output more concise
1. Annotate each test with @org.junit.Test
2. Make methods non-static
3. jh61b.junit.textui.runClasses

example:
https://github.com/Berkeley-CS61B/lectureCode/blob/master/lec4/live1/TestSort.java

Using the "final" keyword in Java

Using the "final" keyword in Java has different meaning depending on what it applied to:

• Variable: the value cannot be changed once initialized.
• Method: the method cannot be overridden by a subclass.
• Class: the class cannot be subclassed.

Notes for "Head first Java"

• A boolean and an integer are not compatible types in Java
• int x = 1;
• While(x)...
• Objects are stored in heap (Garbage-collectible heap).
• If you're running low on memory, the Garbage Collector will run, throw out the unreachable objects (no reference points to them).
• There is no "global variable" in Java. Use public static instead. Any code, in any class of your application, can access a public static method/variable.
• If the end-user doesn't have a JVM, then you'll also need to include that with your application's classes.

[CS61B-2015] Notes for class 3

A variable name in Java is a name for a simple container.
The 9 things that can go in simple containers:

• primitive values(byte, short, int, long, float, double, boolean, char
• references to Objects (an arrow), references may be null

Java is "Pass by value" (in some case, the "value" is a reference?)

[CS61B-2015] Notes for class 2-Defining and using classes

Static vs. Non-static
A class may have a mix of static and non-static members.

• A variable or method defined in a class is also called a member of that class.
• Static members are accessed using class name, e.g. ClassName.StaticMember.
• Non-static members cannot be invoked using class name: Dog.makeNoise().
• Static methods must access instance variables via a specific instance, e.g. an object.
• Static members are shared in a class.

Good Git tutorials

Tutorial
https://berkeley-cs61b.github.io/public_html/materials/lab/lab1/lab1.html#c-using-git

Documentation
http://git-scm.com/documentation

Practice
https://try.github.io/levels/1/challenges/1

[CS61B] Notes for class 5-Linked lists

Array-based lists
Advantage: very fast access of each item
2 disadvantages:
    (1) Insert item at beginning or middle takes time proportional to length of array.
    (2) Array has a fixed length.                        

Linked lists (a recursively data type)
Made up of nodes. Each node has:
(1) an item
(2) a reference to next node in list

Linked lists vs. Array lists
Advantages of linked lists:
(1) Inserting item into middle of linked lists takes constant time, if you have a ref to previous node.
(2) List can keep growing until memory runs out.

Disadvantage of linked lists: Finding n-th item of a linked list takes time proportional to n. (Start at head, walk n-1 nodes.)

The "public" and "private" keywords
Private method or field: invisible and inaccessible to other classes.
Why?
(1) To prevent data form being corrupted by other classes.
(2) You can improve the implementation without causing other classes to fail.


The interface of a class: prototypes for public methods, plus description of their behaviours.

Abstract Data Type (ADT): A class with well-defined interface, but implementation details are hidden from other classes.

Invariant: A fact about a data structure that is always true. "A Date object always stores a valid date." Enforced by allowing access only through method calls.
If we don't have any restriction of a field, than we could declare this filed to "public"; Otherwise, we should declare it as "private."

Not all classes are ADTs! Some classes are data storage units, no invariants; field can be public.

Double-Linked Lists
Insert and delete items at both ends in constant running time.

Sentinel: A special node that does not represent an item. It points to the first and last nodes.

[CS61B] Notes for class 6-Stack frames

The heap stores all objects, including all arrays and all class variables.

The stack stores all local variables, including parameters.

When a method is called, Java creates a stack frame (aka activation record); stores the parameters and local variables.

Stack of tack frame.

Garbage collection only performs on heap.

When a method finishes, its stack frame is erased.

Method "Thread.dumpStack()": Prints a stack trace of the current thread to the standard error stream. This method is used only for debugging.

Parameter Passing
Java passes all parameters by value: copied.
When parameter is a reference, the reference is copied, but the object is shared.

[CS61B] Notes for class 4-Iteration and array

Declare arraies:
int[] a, b[]; //a is 1D array, b is 2D array

Constant:
"final" keyword: value that can never be changed.
BAD: if (month == 2)
GOOD: public final static int FEBRUARY = 2;
              if (month == FEBRUARY)

Output data to csv files

There is a little difference between the code implemented in Python 2 and 3.

[CS61B] Notes for class 3-types, conditionals, iteration and arrays

Primitive types: (don't have to create an object to store it)
byte : 8-bit integer    => -128~127
short:16-bit integer   => -32,768~32,767
int    : 32-bit integer  => -2,147,483,648 ~ 2,147483647
long : 64-bit
float : 32-bit
double: a 64-bit floating-point number
boolean: "true" or "false" ("True" or "False" in Python)
char: a character, 2byte, non-signed

e.g.
long x = 43L;

double & float values must have a decimal point:
double x = 10.0;
float x = 43.9f;

                               Object types                            Primitive types
-------------------------------------------------------------------------------------------
Contains a              reference                                   value
How defined?        class definition                          built into Java
How created?         "new"                                        "6", "3.4", "true", etc.
How initialized?    constructor                                default (usually zero)
How used?            method                                      operators("+", "*",...)


Function: methods declared to return a non-void type.

char c = new char[4];
c[4] = 'x'; //Run-Time error

The size of an array cannot be changed once it has been created.

[ML] Precision / Recall

For classification problem

Precision:
    True positives                            True positives
------------------------   =   ---------------------------------------
#predicted positives         True positives + False positives

Recall:
    True positives                            True positives
------------------------   =   ---------------------------------------
  #actual positives            True positives + False negative


There is a tradeoff between precision and recall.
One way to compute the precision and recall is using F_1 score
                           precision * recall
F_1 score = 2 * -----------------------
                           precision + recall

Reference: https://www.coursera.org/learn/machine-learning/lecture/tKMWX/error-metrics-for-skewed-classes