STAT430代写、R代写、data留学生代做、代写R编程语言调试

日期：2019-05-01 09:59

Final Project

STAT430 Spring 2019

Due date: May 10, 2019

This project aims at applying whatever you have learned from this course and whatever you have learned

from other sources for solving real-world problems using complex financial data. Please read the following

items carefully.

Evaluation

– Your grade on the project will be calculated based on the group options. Please refer to the group

options for the grading details.

Dataset

– The dataset for the final project has to be the futures data provided in Compass 2g. You can use

either the original tick data or the minutes data provided. There will be bonus credits if choose

to preprocess the tick data by yourself. (Update: If you insist on using some alternative high

quality dataset, you have to get my approval in advance.)

– The license of the data only allows students who take STAT430: Machine Learning for Financial

Data to use the data for their homework and/or projects of the course. PLEASE DELETE

THE DATA COMPLETELY AT THE END OF THIS SEMESTER!

Grading components

– Basic required components

create feature matrix and labels

conduct analysis involving at least the following components:

· fully-connected layers

· convolutional neural networks

· recurrent neural networks

· comparisons of at least 3 models (including shallow machine learning models) that are

significantly different

· commonly used regularization methods such as dropout and regularization

the report should include at least the following components:

· title, group number, authors, time

· a short abstract or executive summary

· introduction including but not limited to the objectives of the project

· data description / data preprocessing / sources, etc.

· exploratory analysis

· formal analysis

· conclusion

– Additional effort

The following items will be counted as additional effort:

· adding more relevant features (e.g., technical indicators included in R package TTR can

be looked at additional features)

· significant effort in tuning hyper-parameters and/or architectures so that the candidate

model has non-trivial prediction power

· any other extra effort based on the instructor’s judgment

Project report

– The report should be well organized, and the codes and other technical materials should be put

into appendix.

– The report can be written in either latex, (R)markdown, or word format, and only a pdf file should

be submitted.

– There are no requirements on the number of pages of the report.

Peer review

1

– Only applies for the groups with 3 members

– Peer reviews will be kept strictly confidential in Compass 2g.

– Overall, how efficiently did your group work together on this project.

– Evaluate your peer group members, and assign a grade (out of 100) for their contribution and

performance to the project.

To be submitted

– Group/Individuals: Project report “final-report-[Group number or individual netID].pdf”

– Each member from the groups of 3 members: One-page brief peer review “final-peer-[Your

netID].pdf”

2