How to present our data in different graphs

(A) I have a desired vector u=[1 2 3 4]. I used an algorithm and that algorithm estimated a best solution for my desired vector u. I ran that algorithm 100 times and obtained 100 such estimated vectors of same size as my u vector. That algorithm also estimated 100 values of my fitness function, 100 values of Execution time of that algorithm, 100 values of Absolute Error Vector of same size as u. Then I found the errors in each indivudual element of u and estimated vector. Then I got the mean of 1st two errors separate and mean of last two error separate, thus got 100 such vales for each mean.

(B) Then I added a noise to vector u. and repeated the above process. Then I changed the value of noise and repeated the same process.

(C) Now I increased the number of elements in u from 4 to 6 and repeated the same process with noise and without noise

(D) Now I increased the number of elements in u from 6 to 8 and again repeated step (C).

Now I want to present this data in diffrent such graphs which attract a reader. Can any body help me in this regard?

[Ameer Hamza]

It seems like you want to visualize some statistical results. One way I can think of is to use boxplots or boxcharts

https://www.mathworks.com/help/stats/boxplot.html

https://www.mathworks.com/help/matlab/ref/boxchart.html

Histograms can also be used: https://www.mathworks.com/help/matlab/ref/matlab.graphics.chart.primitive.histogram.html

[Sadiq Akbar]

Thank you very much Ameer Hamza for your guiadance. Yes you are right, but I want to represent my data in such a way that shows that (1) how my algorithm estimated the four values during these 100 iterations (2) how the fitness varies with respect to number of increasing "the no. of elements" in vector u during these 100 iterations (3) What is the % of succes of estimations for 4 elements, 6 elements and 8 elements during these 100 iterations etc. In short to display such statistical different graphs that can show that yes the algorithm is successfull in estimation of these values. Also if you can suggest something else that can show the convergence, robustness etc. of the algorithm.

[Sadiq Akbar]

Dear Ameer Hamza I am attaching one such sheet for you to understand it in a better way. Read the Note in Column K carefully.

Answer 1

[Omega]

Hi Sadiq,

After going through the data attached, I suggest you to consider about the following graphs:

1. Boxplots:

• You can use "boxplot" to show the distribution of the mean squared errors (MSE) for different configurations (4, 6, and 8 elements in u) with and without noise.
• This will help you visualize the variability and outliers in your error data.

2. Bar graph:

• You can use a bar graph to compare the success percentage of the algorithm's estimations for different sizes of the vector u (4, 6, and 8 elements).
• This will help you in evaluating the algorithm's performance and robustness for different vector sizes.

3. Line Graphs:

• Fitness Over Time: You can plot the "fitness" values against "Time (in sec)" to show how the fitness improves over time. This can help illustrate how the algorithm performance changes as the number of elements increases.
• Error Reduction: You can plot the separate line graphs for each error variable (Error in Var1, Var2, etc.) over time to illustrate how errors decrease.

4. Histograms:

• You can use histograms to visualize the distribution of execution times across the 100 iterations for different configurations.
• This will provide insights into the consistency of the algorithm’s execution time.

You can refer to the following documentation link to learn more about the "2-D and 3-D plots"

https://www.mathworks.com/help/matlab/2-and-3d-plots.html