What types of knowledge do you have about the material you are dealing with? How was it collected? When was it collected? Where was it collected? How many people collected it?

You also need to know what type of distribution your data takes.

I have experimental data of compressive strength and permeability for different fly ash based tiles molds

How did you create these molds? Did you use different combinations of fly ash and other materials? Did you use different bake temps and times?

If you use a multiple regression analysis on your data, you will minimize the amount of error. This is why I keep asking pesky questions. It does have a point;-)

Yes at different percentages of fly ash and other materials sand, stone dust, waste polythene fibers, river bed material, along with 20% of cement. I have found at 50% of fly ash with other materials achieved maximum compressive strength and minimum permeability.

Ok. This actually sounds like a Mixture Design. To determine the error properly, you will need some good statistical software like Design Expert or JMP. The method you will use is similar to Multiple Linear Regression. By analyzing all of the things you changed in your test samples at the same time, you will minimize the expressed error in your experiment. The software will determine the error.

You will also have the ability to use the optimization algorithm to do a multi-objective optimization. You can take the models you create for compression strength and permeability and find the most optimal settings for each of the things you tested.

Have you ever used Multiple Linear Regression before?

No I did not use but I will try in this article.

I did experiments on fly ash based roof tiles by falling head method of permeability to assess seepage characteristics of freshly mixed materials. Any one can suggest me which type of graphs I can draw for these results.  

You can calculate it by statistical analysis program.

By using statistical analysis softwares, you can easily calculate the analysis or experimental errors

experimental error =( Actival- theoretical )× 100 divided by Theoretical

You can use artificial neural network methods, such as support vector machine (SVM) or another statistical program.

Working with SVM is much easier than other programs.

Check https://sciencenotes.org/calculate-percent-error/

Also check https://www.thoughtco.com/how-to-calculate-percent-error-609584

Check on this paper.

https://www.thoughtco.com/how-to-calculate-experimental-error-606086

Error Formula

In general, error is the difference between an accepted or theoretical value and an experimental value.

Error = Experimental Value - Known Value

Relative Error Formula

Relative Error = Error / Known Value

Percent Error Formula

% Error = Relative Error x 100%

Example Error Calculations

Let's say a researcher measures the mass of a sample to be 5.51 grams. The actual mass of the sample is known to be 5.80 grams. Calculate the error of the measurement.

Experimental Value = 5.51 grams

Known Value = 5.80 grams

Error = Experimental Value - Known Value

Error = 5.51 g - 5.80 grams

Error = - 0.29 grams

Relative Error = Error / Known Value

Relative Error = - 0.29 g / 5.80 grams

Relative Error = - 0.050

% Error = Relative Error x 100%

% Error = - 0.050 x 100%

% Error = - 5.0%

Dear Mohammad Nadeem Akhtar ,

Use MATLAB or any related statistical analysis software.