Ellen-Wien Augustijn

Ellen-Wien Augustijn

I am a teacher and researcher at ITC. My research focusses on "Spatial Analyses" and "Agent-Based Modeling". I am most interested in spatial-temporal diffusion patterns of infectious diseases.

Location Faculty of ITC, University of Twente, Enschede, The Netherlands


  • Thanks nira, yet, indeed I am very much interested.

  • Dear Nira, thank you very much for your reply. It is very valuable to us that you share your own experiences. Let us hope that the LPG cylinders will make a difference. Will they be provided for free or do families have to pay?

  • I checked the number of participants that have completed the survey yet, it is not possible to create a map based on 18 entries. Perhaps some participants that already completed this week before we fixed the link will still return to this step to complete the survey. If not, we will not be able to create the map.

  • I think you refer to the fact that air pollution is a continuous phenomena (gradual change with distance and can be measured everywhere) yet, infectious diseases are not. You can argue about the level of spatial dependency this type of disease will have. Personally I often think that what you are measuring is the distribution of people.

    There are many...

  • In the model of Shaheen, people can decide to "boil" the water, this would be the same as filters/purifiers as it converts water with infection to clean water.

  • When you want to test out interventions e.g. for diffusion of infectious diseases, you will need to implement agents that decide when to start the intervention (health officials) and what type of intervention to use (isolation, vaccination of all people, or of a particular age group). The timing is difficult because you want to start the intervention before...

  • What do you think of the maps above, showing your use of different types of analytical techniques? Do you think this is realistic? Why/ why not?

  • Yes, I know this example and indeed, finding associations between disease and environmental factors is one of the interesting application domains of machine learning.

  • Yes, you can regard this as an example of machine learning yet, not a spatial example. Did you have the time to read the full article yet it seems to focus an the diagnosis of pneumonia in patients. It is a growing field in diagnosis yet, also in spatial analysis of disease occurrence.

  • Indeed, we discovered that this step was linked to the wrong survey. Our apologies for this. The water survey belongs to week 3. The problem has been resolved yet I am not able to produce a tick bite map for you at the moment. Very sorry about that. Hopefully we will still have a number of people that complete this survey and I will try to generate the map...

  • Yes, the data needs to be spatially aggregated. You can do this at multiple levels in case you are not sure what is the correct level. What you need to do is decide what spatial level is relevant. Maybe the neighborhood is too small and the region is too large, so create a file for all municipalities. Are you interested in both time and space or only space? If...

  • Indeed, it is equally applicable in both domains.

  • Actually R is not that difficult to use. There are however some tricks. R has many packages containing a certain type of functionality. Most of these packages come with an article describing a case study using the package and a manual. Always track the article first and follow the steps described in the article to get a bit familiar with the package. Make sure...

  • Nira, yes indeed you are right.

  • Please take a moment to complete the survey if you wish to contribute to the map of next week. I noticed that the number of participants is still a little low to create the map so I appreciate a few more if possible.

  • You should ask yourself what the functionality of the software should be. You are talking about surveillance data and routine data collection. Do you want to make maps of the data you collect and compare them over time? Or do you need more analytical functionality? Try something open source like QGIS, or use R to plot your maps. Other software may work also.

  • Emilio, are you considering to use data mining to analyze your historic time series?

  • This is a very interesting discussion. We do have a question about expertise ( I the last section of this block). Perhaps more experienced learners give different answers yet, we do not want to exclude anyone. The map does not show the availability of data, it shows if data is felt as a limiting factor. It might even be so that when no data is available at...

  • Indeed, standardizing the way data is entered in a survey can be very important. You could consider asking people to enter their locational information twice (in different ways) for example by placing a dot on a map and by entering a postal code or city name. Just to confirm. Yet, might even take you more time in joining the data.

  • So when we start to use health apps for self diagnostics of many diseases, this opens wide fields of applications. As these apps are spatially aware (gps) these observations are automatically spatial and temporal.

  • Yes, indeed, and this is a very large and growing application domain. Yet, it can also be used for geographic applications.

  • Interesting remark, business intelligence to me seems to be a wider concept, it should be more than just information. But perhaps I do not understand exactly what you mean.

  • Hello Lucy, how fast do you think the increase in volume of data will be? Will this volume double in the coming 10 years? Or do you think the increase will be much faster?

  • The map above has been updated today (November 27). The lower map is the combined group of 2016 and you. This map is based on 240 participants so the results are more stable than the top map which is based on only 77 participants.

  • We have updated the maps above today (Monday November 27). There are some differences between the map of the combined groups (2016 and you) and the result of your data only. However, when more people take part in the survey, this difference might disappear.

  • Thanks for sharing these QGIS tutorials with us. As an institute we use open source software like QGIS a lot more than we used to do. In our distance courses it has been introduced as our standard software.

  • No problem that you cannot provide a paper. It is still good to hear that you reflected on your results, and are still trying to use the knowledge that the data provided you.

  • Those are important points you are raising. In which country did you do your cholera research? Did you publish your work? If so, provide us with a link to your paper. It will be very interesting to read more about your findings.

  • The fact that you are collecting spatial data is already an important step. Did you find the analysis done by the expert useful for your research?

  • Mina, you should keep in mind that the number of respondents for the map was not very large (171), especially when you count per country (although we removed countries with a single respondent). Low numbers of respondents influence the outcome (you might accidently have a sample of people that do have expertise problems).
    This is why it is so important to...

  • There is great potential in large datasets either globally to study for example pandemics, but also more locally (country specific data). I am not sure if there is an immediate risk of not finding essential datasets because of an overload of data. However, quality control and good methods to extract useful information from these datasets are important topics.

  • I always understood that Dengue fever is influenced by very local factors, like roof type and the fact that people grow plants in pots (although one of my students in the course of last year told me she does not believe this - or at least it depends on how the plants are grown and if standing water is left in the pots).

  • Hello Emman, which software will you be using?

  • What type of statistical tools are you using yourself?

  • The map showing the lines is not an animation but the result of a Location-allocation analysis performed in ArcGIS. This offers the possibility to visualize the source a point is assigned to. In case you have access to this software you will find some online tutorials: http://pro.arcgis.com/en/pro-app/help/analysis/networks/location-allocation-tutorial.htm

  • Do you think this is a problem or is it something we should encourage? Do you for example miss out on training (not automatically provided for open source software)? Is the reliability of the software a problem?

  • Narendra Kumar, there is no answer to your question. Which type of task do you want to perform in the software? Do you want to make a map, do analysis? Besides the type of task, it also will depend on your skills, some people like to work in code and prefer for example R, others want a user interface to select menu options and prefer ArcGIS or QGIS.

  • Personally I like to work with multiple software packages, as long as I can load the necessary data from one package to the next without any problems, I prefer to work with the software that is best tuned to the task I want to perform.

  • indeed.

  • When you have a look at the results of your own survey (using the link above) you will see that it is also a problem in other countries.

  • What type of secondary data would be useful to you in your work? Can you give us an example of a type of analysis that is currently not done but would be helpful to you?

  • Nga, I think that these are very valid points you are raising.