Community | Desice - Design of Experiments (DoE)

Q: Feedback, suggestions, missing features and bugs

Welcome at DESICE, we want to keep DESICE as community driven as possible. Therefor we need your feedback! Any suggestions or missing features on the platform? As we know that the platform is far from perfect, bugs can and will happen. If you encounter one, please report them here. Kind regards, Rupert

Asked: May 8, 2021, 8:14 p.m. by Rupert B.

Hi Rupert, Have been trying to do a test run on this platform but I am having trouble choosing factors and responses in the analyse section. Is there a tutorial or readme file to help troubleshooting? Best, John-Paul

Commented: May 25, 2021, 9:43 a.m. by John-Paul A.

Hi John-Paul,
at the moment there is no read-me file available. But the workflow is as follows:

Make sure the data of the experiment is complete and the decimal point is a dot and not a comma
In the analysis section, choose the design you want to analyze
Select the factors and responses (multiple factors can be selected with the ctrl-key) and press the green analyse button --> now you can select the response at the anova dropdown and evaluate the model
Once you got the model you want, the important step is that you save the model to visualize it later on. Select your model in the "Use of model" selection and set your x, y and z axis as well as the constant factor
Press the Visualize button; After that, the response surface should appear in space below; sometimes with linear models the surface appears to be not visible. Then just drag the whole response surface plot and it should become visible

Best regards, Rupert

Commented: May 25, 2021, 7:10 p.m. by Rupert B.

Hi, I've followed these steps but when I hit the visualize button I'm not getting a response. I'm not sure where I should drag from or what step I might've followed incorrectly. Thanks, Kirstie

Commented: Jan. 19, 2022, 5:52 p.m. by Kirstie G.

Q: The analyze module is not working

I cannot generate my sample DOE

Asked: May 9, 2021, 9 p.m. by Carlos V.

Hi Carlos, just looked at your factors names. The problem was, that with two spaces in the factor names, the analyze module could not make the calculations. I added an additional design with underscores at your designs. Let me know if that helped. Kind regards, Rupert

Commented: May 10, 2021, 4:43 p.m. by Rupert B.

Q: Forgotten password function missing

With with current website, it does not appear that the "Forgot Password?" link on the login page works. I'm fortunately logged in right now, but I mistyped my password when making an account and can't get back in now. Suggest to fix the password recovery link, and to add a secondary confirmation password entry box when making new accounts to avoid similar issues in the future. Otherwise, the website seems to function as intended.

Asked: June 1, 2021, 3:51 a.m. by Evan K.

Hi Evan,
the password recovery is not yet fully implemented, but I am working on it. I will post an update, once it is up and running.
Best regards, Rupert

Commented: June 1, 2021, 8:10 p.m. by Rupert B.

Q: Correct path for using the site

First, THANK YOU and THANK YOU! Exactly what I was looking for. (How do I contribute to your fine work?) I have some back ground so the DESIGN was easy to follow and save. I was thinking that ANALYZE would be the tab to add data but I see that the data should already be added. So, if I select EDIT a Design I see that result data can be added. Is this correct? My suggestion is that you help users understand how to get to the screen for adding response data. Maybe another Tab. Most folks will use your DESIGN Tab to design their experiment and then go run it and then add data. Kinda' a step in between Design and Analyze. It would also be nice to print out the design with the Response data field. I chose to edit the design and take a screen shot and print the screen shot but I recommend that you add a function that lets the user print out the design for use in conducting the experiment.

Asked: July 15, 2021, 9:04 p.m. by Charles D M.

Hello Charles,
first of all thank your for your feedback, we are working on streamlining the workflow of the platform. You are right, you add the response data on the edit page of the designs. On this page there is also a button where you can download the experimental plan as an excel-file, so you don't have to take a screenshot for conducting the experiment.
Kind regards,
Rupert

Commented: July 17, 2021, 8:14 a.m. by Rupert B.

Yes, I see the button. But I kinda' like the large format image in your work flow as it is pleasing to the eye and keeps your "brand."

Commented: July 17, 2021, 8:28 p.m. by Charles D M.

Also, do not assume that everyone has or is using Excel. If they do not, then they have to resort to a screen shot. In the spirit of your eZ and self-contained work flow, having a way to print from your program is a nice touch for teh user.

Commented: July 17, 2021, 8:34 p.m. by Charles D M.

Q: I can get the first analyze screen to work but....

I can get the first analyze screen to work but I cannot figure out how to Run the ANOVA and get to interaction analysis. Help Please.

Asked: July 15, 2021, 9:34 p.m. by Charles D M.

Hello Charles,
please refer to the answer to the question of John-Paul on the top. The workflow of the analysis section is described there.
Kind regards, Rupert

Commented: July 17, 2021, 9:45 a.m. by Rupert B.

Please add that the blue Analyze button must be clicked after selecting the experiment. If this button is not clicked, then a response cannot be chose n in the drop down and the user is stuck with no visualization.

Commented: July 17, 2021, 8:39 p.m. by Charles D M.

I made an auto analyze feature, so if you choose a different experiment it gets automatically analyzed.

Commented: July 20, 2021, 6:47 p.m. by Rupert B.

Great feature, thank you!

Commented: July 20, 2021, 10:18 p.m. by Charles D M.

Q: How to delete saved models, etc.

I would like to delete the test models that I create as I try to learn how to use the program. How.

Asked: July 15, 2021, 11:03 p.m. by Charles D M.

Hello Charles,
for the deletion of a model, select the design on top on the page, on which you created the model. On the bottom of the page there is the folder in which the models for this specific design are stored. Once you have that opened, you can delete the model with the delete button.
Best regards, Rupert

Commented: July 17, 2021, 9:52 a.m. by Rupert B.

Yes, that works nicely. Thank You!

Commented: July 17, 2021, 8:32 p.m. by Charles D M.

Q: When I select the Model folder in the ANOVA section...

When I select the Model folder in the ANOVA section I get an error message - Internal Server Error. What am I doing wrong?

Asked: July 15, 2021, 11:10 p.m. by Charles D M.

Hi Charles,
if you don't select a design the platform runs in an error, because it can't find matching models. I fixed the bug, so that don't run in an internal server error.
Best regards, Rupert

Commented: July 17, 2021, 10:48 a.m. by Rupert B.

I understand!

Commented: July 18, 2021, 12:37 a.m. by Charles D M.

Q: I do not see the interaction plots that I am used to and that are "standard"

Still loving your online DOE Tool! The interaction plots that I am seeing in your program are not what I am used to. Usually an interaction plot shows an output (Y) on the vertical axis , One of the factors on the X axis(+ & -) and then the other factor plotted as two lines for when the value is high and the other line when low. It is then possible to see if the factors interact. It appears that you are just plotting the data as if a single factor high and low vs the output and therefore interactions are not displayed. Attached is a link that explains interaction plots. https://www.datasciencecentral.com/profiles/blogs/the-significance-of-interaction-plots-in-statistics -dan

Asked: July 17, 2021, 10:47 p.m. by Charles D M.

Hi Dan,
if you take a look at the interaction plots it exactly does what you described. The only difference to the link you provided is, that the plots are next to each other compared to all interactions in one single plot. The y-axis is scaled to the max values of the response, so you can directly compare the "steepness" of the plot. Best regards, Rupert

Commented: July 19, 2021, 4:55 p.m. by Rupert B.

IMHO. What is provided is basically main effects graphs. You have averaged the Output for one Factor at High and Low Factor Level and have plotted without considering the other factors. This has been done for al of the factors in the three factor design, Yes, their influence is within the data but not specifically comprehended in such a way that interactions can be visualized. I believe each graph you are displaying is a stand alone "Main Effects" graph. I believe that an Interaction Graph comprehends two factors and includes the response values for both of the factors in the averaging. The values that are plotted are the averages of two factors when both are low, when both are high, when one is low and when the other high, and when one is high and the other low. They results must be plotted in the same graph with the x-axis being the high and low values of one of the factors. The two lines represent the other factor at high and low values corresponding to the x-axis conditions. There will be a graph for Factor A vs. Factor B, a graph for Factor A vs Factor C , and a graph for Factor B vs. Factor C. There is an interaction between two independent variables when the effect of one depends on the level of the other and IMHO this is only expressed and visualized using the method I have described.

Commented: July 19, 2021, 11:26 p.m. by Charles D M.

A better graphical representation of the interactions would indeed be helpful. I'll take a look at it.

Commented: July 21, 2021, 8:59 p.m. by Rupert B.

I am glad that you are open for modification. If there is anyway to send interaction math and graphs for the model that you can see in my files, I will be glad to contribute info for your project. Your approach and tools are wonderful and much needed. Thank you, again!

Commented: July 23, 2021, 2:16 p.m. by Charles D M.

Q: Matrix for the Central Composite Rotational Design (CCRD)

Hi! How to insert axial points for a Central Composite Rotational Design (CCRD) type planning? There is a possibility?

Asked: Sept. 30, 2021, 5:57 p.m. by Rubens Henrique L.

Hi Rubens, at the moment it is not possible, but let‘s discuss it via email. Feel free to get in contact at info@desice.io

Commented: Sept. 30, 2021, 8:21 p.m. by Rupert B.

Q: Data link leads to "Internal Error"

I managed to create different experiments, but now the button Data leads me to an "Internal Error" and a blank page.

Asked: Dec. 14, 2021, 7:23 a.m. by Stefan T.

Hi Stefan, please check your E-Mail inbox. Kind regards, Rupert

Commented: Dec. 15, 2021, 6:25 p.m. by Rupert B.

Q: unable to save the design

Ex scope:Parameter screening, Desig: full factorial design Nr. of center points: Zero Factors: three Responses: one

Asked: March 7, 2022, 7:35 a.m. by chethan s.

Hi Chetan, I should work now. Kind regards

Commented: March 7, 2022, 8:36 p.m. by Rupert B.

Q: Delete an experiment

How does one delete an experiment created in error?

Asked: May 1, 2022, 4:09 p.m. by anonymous.

Hello, could you describe your problem in more detail? Basically when you created an experiment, it is possible to delete it again in the "Data" section.
Kind regards, Rupert

Commented: May 2, 2022, 7:44 p.m. by Rupert B.

Q: Visualize button does not work

Hi Rupert! Could you please answer why there are no results after clicking on Visualize button? All setting seems to be correct with no error message.

Asked: May 19, 2022, 12:09 p.m. by Inna S.

Hi Oleksandr, is the chart appearing below the input fields? If yes, sometimes the angle of the view on the 3D-chart is rather unlucky. Try to drag it around in order to see the plane of the response surface. If this doesn't help, feel free to get in contact directly via info@desice.io.

Commented: May 19, 2022, 12:52 p.m. by Rupert B.

Thank you for your support!

Commented: May 21, 2022, 8:31 a.m. by Inna S.

Q: response surface plot scalar scale range

Hi Rupert, Is there a possibility to change the range in scalar scale, sometimes the moving range is automatically set to 2 points, this creates an overlap of range.

Asked: Nov. 14, 2022, 11:19 a.m. by chethan s.

Hello Chetan, at the moment there is no possibility to change the ticks in the colorbar manually. There was a bug which was setting the scale wrong, as it should set it depending on the input range. I fixed it now. Let me know if that helped. Kind regards!

Commented: Nov. 16, 2022, 8:38 p.m. by Rupert B.

Hi Rupert, It works now! Thank you,

Commented: Nov. 17, 2022, 8:32 a.m. by chethan s.

Q: Design of experiments is not working

I am trying to design same DOE as in example and getting error message. Is platform still working? Thanks in advance

Asked: Dec. 15, 2022, 7:50 p.m. by anonymous.

Hi, the platform is still running and the fix is already deployed. We might have been a little bit too strict with special character validation for the units. If you had used f.e. mg_L instead of mg/L it would have worked, but nevertheless you can use special characters in the units again. Kind regards, Rupert

Commented: Dec. 16, 2022, 1:50 p.m. by Rupert B.

Q: Central composite design - high and low settings

Hi, when I choose a central composite design, the experimental design only gives 0, 1, or 0.5 at the experimental design for high and low factor levels. This problem only occours at CCD, not at full factorial or fractional factorial. Could you please fix this bug - thanks :) Daniel

Asked: Jan. 25, 2023, 12:52 p.m. by Daniel G.

Hi Daniel, thanks for the feedback. It should work now. Best regards, Rupert

Commented: Jan. 30, 2023, 8:24 p.m. by Rupert B.

Q: Open source?

Hi Rupert, I was just wondering if you are open sourcing this project? would be great to help out. I think this platform is great and sorely needed. All of the other doe softwares are bloated and slow, not to mention expensive. I work for a cellular agriculture company, and we'd be interested in collaborating to improve it or test it in an industrial environment. Let me know!

Asked: March 6, 2023, 1:29 p.m. by Alex T.

Hi Alex, I sent you an E-Mail. Best regards, Rupert

Commented: March 7, 2023, 7:32 p.m. by Rupert B.

Q: Coefficients of the model terms?

Hi Rupert! I am trying Desice and having a good experience with it. However, how can I see the model parameters (the coefficients of the model terms) after the model building? (I see the ANOVA table and predicted vs actual responses). Thank you very much for this software and for your help!

Asked: April 12, 2023, 11:40 a.m. by Adriano F.

Hi Adriano, thanks for the Info. It somehow went missing in the development. I will include again it in the next release. Kind regards, Rupert

Commented: April 13, 2023, 4:34 p.m. by Rupert B.