Requirements for Understanding and Explaining Your Methodology

1. Understanding the Complete Method

 You must clearly understand the method you are using.
 You should know all steps performed inside the method from start to end.
 If input is given at the start, you must know:
o How the input is processed step-by-step.
o How the processing eventually reaches the output.
 You should be able to explain the full mathematical logic of each step.

2. Preprocessing Stage
 If preprocessing is applied, you must know:
o Which preprocessing techniques are used.
o Why those techniques were selected.
o The mathematics or rules behind each technique.

3. Neural Network Details

If a neural network is used, you must know the following:
a. Architecture
 Which specific neural network model you used.
 Number of layers.
 Number of neurons in each layer.
 Justification for:
o Why you used this number of layers (e.g., 10 layers).
 o Why you did NOT use more (e.g., 15) or fewer (e.g., 6).
 Impact of modifying the architecture:
o What happens if layers are increased or decreased.
o What happens if neurons are increased or decreased.
b. Input and Output Layer
 Number of neurons at the input layer and why.
 Number of neurons at the output layer and why.

4. Activation Functions
You must know:
a. Available Activation Functions
Examples (but not limited to):
 Sigmoid
 ReLU
 Tanh
 Softmax
 Linear
 Leaky ReLU
 Swish
 GELU
 ELU
b. Activation Function Used in Your Project
For the activation function you used:
 Name of the activation function.
  Mathematical expression of the function.
 Why you selected this function.
 What would happen if a different activation function was used.
 Justification for the chosen activation function.

5. Dataset Details
You must know:
 Which dataset you used.
 Number of samples.
 All parameters/features included in the dataset.
 How each feature affects the output.
 Which feature is the most important and why.
o Example: Humidity may strongly influence output.

6. Research Structure
You must be fully aware of:
a. Research Objectives
 What you aimed to achieve.
b. Research Questions
 The questions your study answers.
c. Research Contribution
 Your direct contribution to the field.
 What improvement you introduced.
7. Explaining Your Contribution
 You must clearly explain your contribution in the thesis.
 Example:
o “Our accuracy is 98%. Previous accuracy was 97%.”
 You should explain why accuracy improved:
o Due to preprocessing techniques?
o Due to increasing layers?
o Due to increasing neurons?
o Due to choosing a specific activation function?
This justification is your contribution.

8. Results and State-of-the-Art Comparison

After obtaining your results, you must:
a. Performance Metrics
 Report:
o Accuracy
o Sensitivity
o Delay
o Any other relevant metrics
b. Comparison with Previous Research
 Compare your results with recent papers (2020–2024/2025).
 Comparison must be shown in:
o Graphical form
o Tabular form
  Comparison must clearly show:
o Where accuracy improved
o Where sensitivity increased
o Where delay reduced
 These comparisons must be included in your presentation slides.

You should know the method you are using. You must know which steps are
performed inside the method from the very beginning. If you give input at the
start, then you should know how the input is processed step-by-step and how it
finally reaches the output. You must explain it properly and also explain the
mathematics. For example, at stage 1 if preprocessing is happening, then you
should know which techniques are used inside preprocessing.
And if you say that you applied a neural network, then you should know exactly
which neural network you applied, how many layers it has, how many neurons are
in each layer, and what happens if you increase or decrease the number of layers.
What happens if you increase or decrease the number of neurons. And if you say
you used ten layers, then you must justify why you used exactly ten layers—why
not fifteen, why not six? You must know the justification for keeping the number
of layers at exactly ten or fifteen.
You should also know the number of neurons, how many neurons you applied in
each layer, and why you applied that many. You must know how many neurons
should be present at the input and how many neurons should be present at the
output.
You must know about the activation function what activation functions are
available. All activation functions should be known: for example, sigmoid, ReLU,
tanh, and all others available in machine learning. You should know which
activation function you used in your project, and you should also know its
mathematical expression. You must know the mathematical expression of that
activation function. And you must justify why you used ReLU (or whichever
activation function you used). What would happen if you used some other
activation function instead? You must know the justification.
Besides that, you must know the dataset. Which dataset did you use, what
parameters does it contain, what is the number of samples, and which samples
are included? And for every feature, what is its impact on the output? How many
features are there? For example, if humidity is one of the features, then which
feature is the most important that directly relates to the output? You must know
this.
You must know the contribution of your project what contribution has been
made. This should be clear. The objective of the research, the research questions,
and the research contribution these three things you must know exactly. Mostly
the discussion starts with your contribution. In your thesis you should state the
contribution what you contributed that improved the results.
When questioned, you can say, “Our accuracy is 98%; the previous one was 97%.”
Then the question arises: why did the accuracy improve from 97% to 98%? What
caused the improvement? That reason is your contribution. You should say, “We
applied preprocessing in this way,” or “We increased the number of layers,” or
“We increased the number of neurons,” which improved the accuracy.
And in the results section, you must compare with the state-of-the-art. After
finding your results suppose you say you achieved 98% accuracy, this sensitivity,
this delay then you must compare with previously published research papers from
2020 to 2024 or 2025. You must compare both in graph form and in tabular form.
It is mandatory to include these in your slides. And while comparing with previous
research, you must clearly mention what improved whether accuracy improved,
sensitivity increased, or delay decreased. These must be clearly mentioned.
Here is the same content rewritten again, clean, organized, and professional, with
bullets and proper formatting:

Complete Guidelines for Explaining Your Methodology

1. Understanding the Method
 You must fully understand the method you are using.
 You should know every step that occurs inside the method from the very
beginning.
 When input is given, you must know:
o How the input is processed step-by-step.
o How it finally produces the output.
 You must explain each step clearly, including the mathematical operations.
2. Preprocessing
 If preprocessing is applied, you must know:
o Which preprocessing techniques are used.
o Why those techniques were selected.
o What mathematical or logical operations each technique performs.

3. Neural Network Explanation

a. Model Architecture
You must know:
 Which neural network model you used.
 Total number of layers.
  Number of neurons in each layer.
 Justification for the number of layers:
o Why you used exactly 10 layers.
o Why not 15 or 6.
 Effects of changing architecture:
o If you increase or decrease layers → what impact does it have?
o If you increase or decrease neurons → what changes occur?
b. Input and Output Neurons
 Number of neurons in the input layer and why.
 Number of neurons in the output layer and why.

4. Activation Functions
You must know:
a. Types of Activation Functions
Common activation functions include:
 Sigmoid
 ReLU
 Tanh
 Softmax
 Leaky ReLU
 Linear
 ELU
 GELU
 Swish
b. Activation Function Used in Your Project
For the activation function you selected:
 Name of the activation function.
 Its mathematical formula.
 Why you used it.
 What would happen if a different activation function were used.
 Complete justification for your choice.

5. Dataset Knowledge
You must clearly know:
 Which dataset you used.
 Total number of samples.
 All features (parameters) included in the dataset.
 How each feature affects the output.
 Which feature is the most important.
o Example: If humidity is a feature, how strongly it affects the output.

6. Research Foundation
You must be clear about:
a. Research Objectives
 What you aimed to achieve.
b. Research Questions
 What questions the research attempts to answer.
c. Research Contribution
  What improvement or innovation you added.
 What change in your approach led to better results.

7. Explaining Your Contribution

 You must highlight your contribution in the thesis.
 Example explanation:
o “Our model achieved 98% accuracy. Previous studies achieved 97%.”
 You must explain the reason behind this improvement:
o Better preprocessing?
o More layers?
o More neurons?
o Different activation function?
Your justification = your contribution.

8. Results and State-of-the-Art Comparison

a. Performance Reporting
You must provide:
 Accuracy
 Sensitivity
 Delay
 Any other important metrics
b. Comparison with Previous Studies (2020–2024/2025)
 Compare your results with recent research papers.
 Present the comparison in:
 o Graph form
o Table form
c. Highlight Improvements
You must clearly show:
 Where accuracy improved
 Where sensitivity increased
 Where delay decreased
 All improvements must be stated clearly in your slides and thesis.