19th Ave New York, NY 95822, USA

# 数学代写|QBUS6850 Machine Learning

## QBUS6850课程简介

Machine Learning is a fundamental aspect of data analytics that automates analytical model building in modern business. In the big data era, managers are able to use very large and rich data sources and to make business decisions based on quantitative data analysis. Machine Learning covers a range of state-of-the-art methods/algorithms that iteratively learn from data, allowing computers to find hidden patterns and relationships in such data so as to support business decisions. This unit introduces modern machine learning techniques and builds skills in using data for everyday business decision making. Topics include: Machine Learning Foundation; Modern Regression Methods; Advanced Classification Techniques; Latent Variable Models; Support Vector Machines (SVM) and Kernel Methods; Artificial Neural Networks; Deep Learning; and Machine Learning for Big Data. Emphasis is placed on applications involving the analysis of business data. Students will practise applying machine learning algorithms to real-world datasets by using an appropriate computing package.

## Prerequisites

At the completion of this unit, you should be able to:

• LO1. differentiate different types of learning algorithms and identify the advantages and limitations of each method
• LO2. build a strong machine learning skill set for business decision making
• LO3. create machine learning models for studying relationship amongst business variables
• LO4. work with various data sets and identify problems within real-world constraints
• LO5. demonstrate proficiency in the use of statistical software, e.g. Python, for machine learning model implementation
• LO6. work productively and collaboratively in a team
• LO7. present and write insights and suggestions effectively, professionally and ethically.

## QBUS6850 Machine Learning HELP（EXAM HELP， ONLINE TUTOR）

$\left({ }^*\right)$ Set $x_0=10$ to every $\mathbf{x}_n$ instead of $x_0=1$, and repeat the 1000 experiments above. What is the median number of updates before the algorithm returns W PLA $^{\text {? }}$ Choose the closest value.
[a] 8
[b] 11
[c] 14
[d] 17
[e] 20

(*) Set $x_0=0$ to every $\mathbf{x}n$ instead of $x_0=1$. This equivalently means not adding any $x_0$, and you will get a separating hyperplane that passes the origin. Repeat the 1000 experiments above. What is the median number of updates before the algorithm returns $\mathbf{w}{\mathrm{PLA}}$ ?
[a] 8
[b] 11
[c] 14
[d] 17
[e] 20

$\left({ }^*\right)$ Now, in addition to setting $x_0=0$ to every $\mathbf{x}n$, scale down each $\mathbf{x}_n$ by 4 . Repeat the 1000 experiments above. What is the median number of updates before the algorithm returns $\mathbf{W}{\text {PLA }}$ ? Choose the closest value.
[a] 8
[b] 11
[c] 14
[d] 17
[e] 20

Which of the following set of $\mathbf{x} \in \mathbb{R}^3$ can be shattered by the $3 \mathrm{D}$ perceptron hypothesis set? The set contains all hyperplanes of the form with our usual notation of $x_0=1$ :
$$h_{\mathbf{w}}(\mathbf{x})=\operatorname{sign}\left(\sum_{i=0}^3 w_i x_i\right) .$$
[a] ${(7,8,9),(17,18,19),(27,28,29)}$
[b] ${(1,1,1),(7,8,9),(15,16,17),(21,23,25)}$
[c] ${(1,1,3),(7,8,9),(15,16,17),(21,23,25)}$
[d] ${(1,3,5),(7,8,9),(15,16,17),(21,23,25)}$
[e] ${(1,2,3),(4,5,6),(7,8,9),(15,16,17),(21,23,25)}$