讲解QBUS6840、辅导Canvas留学生、Python编程语言调试、Python辅导

日期：2019-09-10 11:27

Predictive Analytics

Semester 2, 2019

Homework (15%)

1 Rationale

This assignment is designed to help students to develop basic predictive analytics skills on synthetic

and possible real applied problems. The skills include data visualization, model building

and analysis in terms of understanding the theory, practicing with raw data and programming in

Python.

2 Questions

The data set SP500 weekly Jan1988 Nov2018.csv (downloaded from https://au.finance.yahoo.com),

available on Canvas, contains the Standard & Poor weekly adjusted closing stock indexes from

January 1988 to November 2018. Denote this time series as {yt, t = 1, ..., T = 1613}.

(a) Write Python script to load the data and produce their time series plot. Include the plot

and your comments (if any) together with the Python script in your submission.

(b) Write Python script to produce one-step-ahead forecasts for the last 100 observations using

the naive forecasting method with drift, i.e. compute

y?t|t≠1, t = 1514, ...1613.

? Plot these forecasts together with the actual values. Include the plot and the Python

script in your submission.

? Report the scale-dependent measure Root Mean Squared Error (RMSE) and scaleindependent

measure Mean Absolute Percentage Error (MAPE) (the errors between

forecasts and the ground true indexes].

(c) Smooth the time series using centered MA-4. Plot the smoothed time series together with

the original time series. Include the plot and the Python script in your submission.

(d) Given the stock indexes {yt, t = 1, ...T}, the stock returns are defined as

rt = log yt+1

yt

, t = 1, ...T ≠ 1.

Write Python script to compute the stock returns and produce their time series plot. Comment

on this plot in conjunction with the plot of the indexes {yt, t = 1, ..., T}. Which dataset

do you think is more predictable, and why? Include the plot and the Python script in your

submission.

(e) For the index dataset {yt, t = 1, ...T}

? Use the last 100 observations as testing data, and the previous observations for the

training data. Use the training dataset to estimate the parameters (weight – and initial

level l0. You may set l0 = y1 or l0 be the average of a few first observations) of the

Simple Exponential Smoothing (SES) method.

? Based on these estimates of – and l0, compute one-ahead-forecasts on the test data, i.e.,

y?t|t≠1, t = 1514, ...1613.

Compute the Mean Absolute Percentage Error (MAPE) and plot the forecasts. Please

also include your Python code in submission.

(f) For the squared returns dataset {xt = r2

t , t = 1, ...T ≠ 1}

? Use the last 100 observations as testing data, and the previous observations for the

training data. Use the training dataset to estimate the parameters (weight – and initial

level l0. You may set l0 = x1 or the average of a few first observations) of the SES

method.

? Based on these estimates of – and l0, compute one-ahead-forecasts on the test data, i.e.,

x?t|t≠1, t = 1513, ...1612.

Compute the Mean Absolute Percentage Error (MAPE) and plot the forecasts. Please

also include your Python code in submission.

(g) Compare the MAPE obtained for the index dataset {yt, t = 1, ...} and the MAPE for the

squared returns dataset {xt = r2

t , t = 1, ...}. Give your comments

(h) (Optional) Repeat parts (e) and (f) with the Trend Corrected Exponential Smoothing (TCES)

method. Do you obtain better forecasts (compared to the SES method)?

Page 2

3 Instructions and Marking criteria

For parts (e) and (f), the more accurate forecasts (i.e., smaller MAPE) you have, the more marks

you’re given. Part (h) is optional, bonus marks might be awarded if you attempt this part.

Assignment Report: The assignment report should be presented as a technical report that:

? details ALL required steps.

? provides sucient

explanation and interpretation of any results you obtain. Output without

reasonable justifications will not receive full marks.

? Clearly and appropriately presents any relevant tables, graphs and screen dumps from programs

if any. You may insert small sections of your code into the report for better interpretation

when necessary. Find the best and most structured way to present your work,

summarise the implementation procedures, support your results/findings and prove the originality

of your work.

? reports numbers with decimals to the three-decimal point.

? properly cites all the references if any.

Assessment of your written presentation skill is part of this assignment. Markers will allocate up

to 10% of the mark for presentation.

Important notes:

? Required submissions: ONE written report (MS word or pdf format) and ONE Python

source code file (Jupyter Notebook .ipynb or .py). Please follow instructions for submissions

announced on Canvas.

? The late penalty for the assignment is 5% of the assigned mark per day, starting after 4:00pm

on the due date. The closing date is the last date on which an assessment will be accepted

for marking. See Canvas for the due date and closing date of this assignment.

? As anonymous marking policy, only include your Students ID in the report and do NOT

include your name.

? The name of the report and code file must follow the format

<your SID> QBUS6840 Assignment1 2019S2.

? The report should be NOT more than 15 pages including everything like text, figure,

tables and small sections of inserted codes, etc, but excluding the appendix.

Page 3

? The University of Sydney takes plagiarism very seriously. Please be warned that plagiarism

between individuals is always obvious to the markers and can be easily detected by Turnitin.

Key rules:

? Carefully read the requirements for each part of the assignment.

? Please follow any further instructions announced on Canvas, particularly for submissions.

? You must use Python for the assignment. To avoid any potential issues with your codes,

please use the latest Anaconda version for your Python programs.

? Reproducibility is fundamental in data analysis, so that you are required to submit a code

file that generates your results. Not submitting your code or submitting code that are not

runnable will lead to a loss of 50% of the assignment marks.

? Referencing: Harvard Referencing System.

(Please find the details at: http://libguides.library.usyd.edu.au/c.php?g=508212&p=3476130)

Page 4