# BEST代写-线上编程学术专家

Best代写-最专业靠谱代写IT | CS | 留学生作业 | 编程代写Java | Python |C/C++ | PHP | Matlab | Assignment Project Homework代写

# C++代写 | PHAS0100: Research Computing with C++

### C++代写 | PHAS0100: Research Computing with C++

PHAS0100: Research Computing with C++
Assignment 1: Constructing a Small Research Project

Part A: Linear Regression App (55 marks)
The first part of this coursework is to get you to setup the project. These instructions will guide you
through.
1. Please read chapter 4 of the “Hands on Machine Learning” that covers linear regression and
provides background to the equations used in this assignment. The PDF is on Moodle in the
Coursework topic.
[0 marks]
2. Data can come from many places, e.g. file, network, or randomly generated. So, we first
define an interface of what we expect from our data provider. In class we learnt: “program to
interfaces”, so:
a. Create a header file, containing a pure virtual interface class, with a method
equivalent to:
virtual std::vector<std::pair<double, double> > GetData() = 0;
The data returned should be a vector of X, y pairs, where X is the observed feature
value, and y is the target/label/predicted value.
b. Ensure the file is included in CMakeLists.txt, so that your build environment will know
it exists.
c. Create a header and implementation file of a new concrete (i.e. not abstract) class
that implements this interface. At first, just write an empty method with the signature
above.
d. Create a unit test file, that will instantiate an instance of your new concrete class.
e. Check that you can compile and run the test.
3. Now we implement the class to generate some data. The idea is that if we create some fake
data, we know what the answer should be.
a. In the class that you created as part of 2c, implement a function that generates data
that fits the linear model: � = �!� + �” + �����
b. In class we learnt the RAII pattern and dependency injection pattern, rather than using
setters/getters. Ensure that parameters for your generator are passed in via
constructor.
c. Write a specific unit test that checks:
i. The number of returned items is correct
ii. The distribution of the returned items is correct.
4. Similar to part 2, create a pure abstract interface for the solver, and a concrete
implementation. Notice how we have separated the thing that generates or provides data from
the thing that provides a solution.
a. Create a header file, containing a pure virtual method to fit data to a model. For this
simple exercise, we know that the model only requires 2 parameters, �” and �!, so the
returned value can be a pair of doubles representing �” and �!.
i.e. equivalent to:
std::pair<double, double> FitData(std::vector<std::pair<double, double> >)

c. Create a header and implementation file of a new concrete (i.e. not abstract) class
that implements this interface. At first, just write an empty method.
d. For simplicity re-use the unit test file created in section2d. (0 marks)
e. Check you can compile and run the test.
5. Implement a first solver, using Normal Equations. See equation 4-4 in “Hands on Machine
Learning” and the Notations section in Chapter 2. This would be placed in the concrete class
created in part 4c.
Hints: This project includes Eigen. Copy data from STL arrays to Eigen, and solve. Don’t
[Implementation 5 marks, Unit tests 5 marks. Both at markers discretion, 10 marks total]
6. Implement a second solver. As mentioned in class the point here is to demonstrate how 2
methods can co-exist in a project, and the project should be able to run both of them.
a. Create a new solver using Gradient Descent.
b. Ensure the parameters are all adjustable, by whoever is using the class (see
Dependency Injection covered in lecture 4).
7. Now, in reality, you would be processing data produced by some scientific experiment. The
idea here is to now create another concrete implementation of the interface defined in section
2a, where instead of randomly generated data, data is read in from a text file.
a. Create a new header and cpp file, of a concrete class that implements your interface
containing the GetData() method.
b. Implement the class, using STL funtions to read data from a plain text file. Assume 2
values per line, representing X and y, each space separated.
c. Write unit tests to ensure you can read TestData1.txt and TestData2.txt (provided) 