I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Others will have to comment on this, but what we should do anyway is to add more colours to the palette to allow for a combination of the two - fixed class breaks (predefined legend set) and that nice colour/value gradation Bob mentioned.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Others will have to comment on this, but what we should do anyway is to add more colours to the palette to allow for a combination of the two - fixed class breaks (predefined legend set) and that nice colour/value gradation Bob mentioned.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Others will have to comment on this, but what we should do anyway is to add more colours to the palette to allow for a combination of the two - fixed class breaks (predefined legend set) and that nice colour/value gradation Bob mentioned.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
I agree with Bob and since the topic is on the table… I´ve been wondering for a long time now…
Why the default legend is painting in green the high values and red the low ones?
I believe that it may apply in some cases, but generally… it ends up in showing areas high incidence of cases in green… and low ones in red… which is not very intuitive for me…
Although…it could be solved if we don´t mix primary colors any more
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Marta, my only thought is that in terms of performance indicators, higher data points would indicate better performance, while lower points indicate poor performance…
I agree with Bob and since the topic is on the table… I´ve been wondering for a long time now…
Why the default legend is painting in green the high values and red the low ones?
I believe that it may apply in some cases, but generally… it ends up in showing areas high incidence of cases in green… and low ones in red… which is not very intuitive for me…
Although…it could be solved if we don´t mix primary colors any more
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Yes Nicola, you are right… it is a good combination for that case… also for high coverage of vaccination… for instance.
I was thinking more in … high number of cases of …, high rate of incidence of…, that generally are not very positive
···
On 20 February 2014 17:19, Nicola Hobby nhobby@psi.org wrote:
Marta, my only thought is that in terms of performance indicators, higher data points would indicate better performance, while lower points indicate poor performance…
I agree with Bob and since the topic is on the table… I´ve been wondering for a long time now…
Why the default legend is painting in green the high values and red the low ones?
I believe that it may apply in some cases, but generally… it ends up in showing areas high incidence of cases in green… and low ones in red… which is not very intuitive for me…
Although…it could be solved if we don´t mix primary colors any more
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
On 20 February 2014 17:19, Nicola Hobby nhobby@psi.org wrote:
Marta, my only thought is that in terms of performance indicators, higher data points would indicate better performance, while lower points indicate poor performance…
I agree with Bob and since the topic is on the table… I´ve been wondering for a long time now…
Why the default legend is painting in green the high values and red the low ones?
I believe that it may apply in some cases, but generally… it ends up in showing areas high incidence of cases in green… and low ones in red… which is not very intuitive for me…
Although…it could be solved if we don´t mix primary colors any more
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Interesting point about the red-yellow-green combo. These are the traffic light colours which are apparently sufficiently ubiquitous that you can say red=bad and green=good without even explaining further. If snooker was as well known as traffic lights we could do a whole range from red through to pink
But in general, using hues as per Knut’s colour brewer link is going to be the best bet for most cases of numeric data. You can even print to black and white printers and maintain the message.
Good point about “equal counts”. I actually did do that but somehow it rendered different on the dashboard.
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
Interesting point about the red-yellow-green combo. These are the traffic light colours which are apparently sufficiently ubiquitous that you can say red=bad and green=good without even explaining further. If snooker was as well known as traffic lights we could do a whole range from red through to pink
But in general, using hues as per Knut’s colour brewer link is going to be the best bet for most cases of numeric data. You can even print to black and white printers and maintain the message.
Good point about “equal counts”. I actually did do that but somehow it rendered different on the dashboard.
I don’t think there’s one single rule that will fit all when it comes to choosing your symbols or colours for a thematic map depicting numeric values. There’s nothing wrong with say a red/yellow/green colour scheme if you are showing for example reporting rates, where low (red) = bad and high (green) = good. If I were to map out HIV prevalence, then I may go from green (low values) to red (high prevalence). More “neutral” variables such as population density are usually better off with a graduated scheme from light to dark, e.g. pale to dark brown.
Maybe the only basic rules are that there should be some logic in the choice of colour ramps, and to keep it simple (don’t overload your map, don’t use 15 different classes each with a different colour when you only have 20 districts!). I must agree that the current colour scheme for ANC1 and ANC2 coverage by chiefdom in the demo site can definitely be improved: too many colours that are not in a logical order. Your diarrhea map looks better, though it would be even better if you changed the classification type to “equal counts” rather than “equal intervals”.
Agree with what has been requested recently by Knut: option to display polygon data (e.g district) by symbol, not just colour, and include more symbols because currently it’s rather limited. Especially for categorical values the legend behaviour could be greatly improved. Bar/Pie charts as symbols on the map would be cool but not a priority as there are more suitable tools for that.
I am no expert here, but I was talking last week with someone who had just been on a GIS course. She made the interesting observation that when you are displaying numerical data as shaded regions on a map, you should not mix primary colours in the same gradated scale because what you end up with is sort of rubic cube kaleidoscope of colour rather than a clear gradation of values.
If that is good advice then I think our demo site could be improved as presently it seems like a good example of bad practice, where we have mixed red, green and blue with joyful and cheerful abandon
I had a go to change the Diarrhoea map on the dashboard at http://apps.dhis2.org/dev/dhis-web-dashboard-integration/index.action****and I think it is true that it conveys the data better than the extreme multi-coloured ones around it. Even if the distribution of data between bands is not so interesting. Anyone agree or disagree? (the data will only be there for a short while longer)
If there is a best practice to be had here regarding colour selection I guess we should promote it, particularly with default offerings so that people have to be quite deliberate in order to go astray.
