CA review - comment #1

[whitere123]

Hello Malavika,

I have reviewed your CA. I will present my comments about your CA as issues on your GitHub.

1:

I think that you should add a mathematical notation subsection in the reference material. On the SRS we have a section 2.4 for mathematical notation. It would be more appropriate to put the summation definition , convention for functions, matrix conventions and definition of matrix operations in this section. For example, DD6 could go into the mathematical notation subsection. Also you could elaborate more on things like summation notation; for example Sigma = x1 + x2 + …. + xn.

[Malavika-Srinivasan]

@whitere123 Hi Robert,

Thank you for your feedback. I appreciate it.

1. I like your idea of having a subsection for mathematical notation but it is adding something to the template which will require approval from Dr.Smith(@smiths ). I will wait for his reply in this regard.
2. I can see what you mean when you say adding definitions for Sigma, matrix operations etc but wouldn't that be too basic? Because the intended reader needs to be proficient in undergraduate level mathematics, I am thinking it would be adding information which is too much basic for them. Just a thought.

Thanks, Malavika

[whitere123]

OK. I think that you may want to at least mention at some point what the rows and columns of the matrix mean (in words) with respect to the given data set. (ie, each row is a new data point and each column is a descriptor of that point).

[smiths]

@Malavika-Srinivasan, I'm fine with a subsection that describes the mathematical notation. As @whitere123 suggested, this would belong with the reference material.

As far as defining mathematical notation, summation, matrix operations, etc, you almost can't err on the side of giving too much information. Defining the meaning of the entries in the matrix (rows and columns) is a great point.

Thank you for your comments @whitere123.

[Malavika-Srinivasan]

@whitere123 Hi Robert,

Can you please guide mea little more on how to phrase this in mathematical terms. I know how to construct the matrix from a system of linear equations or from data points. But I don't understand the term "each column is a descriptor of that point".

What exactly you mean by descriptor, I think each column lists the coefficient of the same degree from the system of equations. Please bear with my ignorance.

Thanks,

Malavika

[whitere123]

"descriptor of that point" was bad choice of words. I will be more clear. That was my mistake.

You should show step by step how you arrived at some of the matrix systems. For example, for the monomial model you (1) assume that the data fits an n-1 degree polynomial. (2) plug each data point into the assumed model and arrive at n linear equations (show them) and than (3) write it in matrix form (ie a row of A dot x is equal to y). I think you should write out these steps as a derivation or validation of the instance models. In the SRS, dr smith asked us to show theory of equations and derivations of models/systems under the instance model tables.

When applicable, you may also want to explicitly relate A to the basis functions (you do this for DD3 but not DD1). For DD4 you may want to take the derivative of the basis functions and plug in the data set to show how you arrived at A.

I just think it should be a bit more clear how the data, basis functions and matrices are related.

[Malavika-Srinivasan]

@whitere123

Yes Robert. That will make it more clear. But I don't want the document to be so detailed. Because I already mentioned in the characteristics of the reader that they need to be knowledgable in these techniques. It is like teaching them how equations are formed and how to solve them using matrix which is not the main purpose of this project. I think that can itself be another project which can show how to solve system of linear equations using different methods and metrics may be one of the method, I am not sure. Here I am just using the idea assuming they have this knowledge. The reader should be proficient enough to know what basis functions are and how they can be solved using matrices.

You are correct that Dr.smith asked us to show the derivation but that applies to something which is derived using basic science. For eg: my Master's project uses law of thermodynamics, newtons law etc which I don't derive and show. Whereas one of my IM is derived because it is something that we derive for our problem and not just the basic science. Here I just use basic math taught to anyone who knows system of linear equations. I think I don't have to teach them how `n' equations are constructed and how they can be put in the matrix.

[whitere123]

Alright, maybe it would be providing too much detail. This is just my personal preference. My philosophy is to try and fill in as much theoretical detail as possible, even if it may seem redundant/tedious , so that I am confident that the reader will know exactly the role each part plays in the software. Your document would still read fine without this extra detail.

Robert

[Malavika-Srinivasan]

@whitere123 ,

No Robert thats a god point. But its just that I felt giving too much information is not necessary as it is a pre-requisite for reading this document. I am closing this issue.