## 这是一篇来自美国的关于机器学习项目1：寻找多项式的代写

**Overview **

This assignment is meant to give you practical experience with machine learning style programming.

At an engineering level, you will need to learn to work with datasets, commandline arguments, file paths, and numerical programming. At an academic level, you will deepen your understanding of vectors, matrices, loss functions, overfitting vs generalization, and more broadly speaking, the fundamental principles of supervised learning.

Your objective is to implement equation 3.28 from PRML – i.e., to find the best fitting regularized weights for a linear model. This part should be straightforward – you *must use python* and either numpy or pytorch. The instructor is more familiar with numpy,so if you choose pytorch, you are in uncharted territory. For the rest of this document anywhere you see numpy feel free to replace with pytorch. In either case, your program must be *vectorized* and reasonably efficient. *You may not do any slow single threaded pure python numerical computation loops*. E.g., you should use the vector manipulation routines from numpy. These are implemented in compiled languages which are much more efficient than python, and if you use python for ML, you need to know how to do it efficiently. You should implement equation 3.28 somewhere in your code as a method.

You will be recreating the results on polynomial fitting from the first chapter. I have created various synthetic datasets, each one being created by choosing a particular polynomial,sampling a subset of the x-axis, evaluating the polynomial at each point, and then adding a small amount of zero-mean Gaussian noise. Your objective is to identify for each dataset the best fitting polynomial degree that does not yield substantial overfitting.

You should write an additional method that sweeps through each polynomial order (up to a given maximum). For each step, you should then find the best weights and evaluate accuracy (via rmse) and estimate the degree of overfitting. You will then have to develop your own heuristics to identify the ”best” order as the one with the highest accuracy achievable without significant overfitting. That is the primary goal of the assignment.

Additionally, you must include a readme with a brief discussion of your approach and the results of your method. **This is an individual assignment and all work must be ****your own. **

**Project Requirements **

This section details the project requirements in terms of the required API, file name conventions, documentation requirements (a report style readme), and data file formats.

**API **

In order to facilitate automated testing, you must name your program polyhunt.py (or something similar, if you use Java or C) and your readme should be named readme.txt.

Since everyone will be having the same filenames, it is critical that you include your names and UR email’s in the contents of every file, and implement the ”info” option which prints your name and contact information. You must submit a program which can be operated via the command line. No Jupyter notebooks. **You must include a readme which ****explains how to use your program and enable/disable the various options. You ****are not allowed to require the TAs to edit your code to change the options. **

Your program must support the following commandline arguments:

- m – integer – polynomial order (or maximum in autofit mode)
- gamma – float – regularization constant (use a default of 0)
- trainPath – string – a filepath to the training data
- modelOutput – string – a filepath where the best fit parameters will be saved, if this is not supplied, then you do not have to output any model parameters

- autofit – boolean – a flag which when supplied engages the order sweeping loop, when this flag is false (or not supplied) you should simply fit a polynomial of the given order and parameters. In either case, save the best fit model to the file specified by the modelOutput path, and print the RMSE/order information to the screen for the TA to read.

- info – boolean – if this flag is set, the program should print your name and contact information

You may define additional optional arguments of your own choosing. Some suggestions:

- numFolds – the number of folds to use for cross validation
- devPath – a path to a held-out data set

**程序代写代做C/C++/JAVA/安卓/PYTHON/留学生/PHP/APP开发/MATLAB**

本网站支持淘宝 支付宝 微信支付 paypal等等交易。如果不放心可以用淘宝交易！

**E-mail:** itcsdx@outlook.com **微信:**itcsdx

如果您使用手机请先保存二维码，微信识别。如果用电脑，直接掏出手机果断扫描。