The ultimate Fabric, Power BI, SQL, and AI community-led learning event. Save €200 with code FABCOMM.
Get registeredEnhance your career with this limited time 50% discount on Fabric and Power BI exams. Ends August 31st. Request your voucher.
Hi, need some help on what I am doing wrong here.
The end-goal I wish to achive is a distribution curve combined with histogram.
I have been looking into the use of density in deneb but do not seem to get it right.
Code below
{
"data": {"name": "dataset"},
"layer": [
{
"description": "Histogram Bars",
"mark": {
"type": "bar",
"color": "#7DA7D9",
"tooltip": true
},
"encoding": {
"x": {
"bin": {"maxbins": 40},
"field": "Diameter",
"type": "quantitative",
"scale": {
"domain": [7.82, 7.87]
},
"axis": true,
"title": ""
},
"y": {
"aggregate": "count",
"type": "quantitative",
"axis": {"labels": false},
"title": "Count"
}
}
},
{
"description": "Histogram line",
"transform": [
{
"density": "Diameter",
"bandwidth": 0.3
}
],
"mark": {
"type": "line",
"color": "#931313",
"strokeWidth": 2
},
"encoding": {
"x": {
"field": "value",
"type": "quantitative"
},
"y": {
"field": "density",
"type": "quantitative"
}
}
}
],
"resolve": {
"scale": {"y": "independent"}
}
}
Any ideas where the problem is?
Solved! Go to Solution.
@KerKol, I adjusted the bandwith to match the standard deviation for the data sample used and got the following result.
@KerKol wrote:sounds like you may have found the issue, the above curve looks about right.
You have the Y aggregated as a count for your bars, but perhaps not for your line. What happens if you adjust the aggregation?
For the y-axis of the bars I use the aggregate: "count" and for y-axis I use "field": "density".
I am curious about what you mean by "adjust the aggregation".
The code works for me on my own data. What happens when you set resolve to dependent?
And remove the domain scale?
Proud to be a Super User!
@KerKol thanks for your suggestions but it did not help...
When I set to dependent it moves sligthly along the x-axis.
When I also remove the domain scale the bar streches out along the x-axis. In both cases the line is still not changing.
How about the "x": {"field" : "value"}... What is really inside "value", how can its content be verified or changed?
I tried changing the bandwith since Diameter has several decimals and started to get some effect..
sounds like you may have found the issue, the above curve looks about right.
You have the Y aggregated as a count for your bars, but perhaps not for your line. What happens if you adjust the aggregation?
Proud to be a Super User!
@KerKol, I adjusted the bandwith to match the standard deviation for the data sample used and got the following result.
@KerKol wrote:sounds like you may have found the issue, the above curve looks about right.
You have the Y aggregated as a count for your bars, but perhaps not for your line. What happens if you adjust the aggregation?
For the y-axis of the bars I use the aggregate: "count" and for y-axis I use "field": "density".
I am curious about what you mean by "adjust the aggregation".
Привет, мне так же интересна эта тема. Разместите полный код JSON, спасибо!