Hi,
I am currently trying to create a Stacked Area Chart. The Problem is, that it creates gaps on several positions. i already red in other threads, that this happen, if the different categories of the data doesnt start on the same time step and you can add 0 to prevent this, but my Data is complete. The problem seems to be the interpolation of the data. Here is the Code:
p=ggplot(Df_plot, aes(x=Time, y=Values, fill=Tech)) +
it creates:
Here is an example of the Data:
"Time" "Tech" "Values"
Hope someone can help!
If you think about it, area is continuous, and your data is discrete -- not every event happens every day. I think you'll do better with geom_col
(which is all I could do with the sample data).
ggplot(data = series, aes(x=Time, y = Values)) + geom_col()
karawoo
December 8, 2018, 12:29am
3
The problem you are seeing is due to having a mix of positive and negative values in some series. There is no real way to avoid the gaps as the areas switch back and forth from positive to negative. There is some more discussion of this on this ggplot2 issue . geom_col() might indeed be your best bet here.
1 Like
thanks for the Answers...with the separation of the positive an negative values i could erase the gaps in the plot.
The issue now is that (probably) due to the interpolation of these data, there are positive and negative Values at the same Time. Does someone know what to do?
Here is an example of the data:
Df_plot = data.frame(Time,Tech,Values)
p=ggplot(Df_plot) +
"Time" "Tech" "Values" "positive" "negative"
That's good fodder. So help me out, please.
Time is a sequential vector of observations, right? Tech is a category. Values measure some attribute of Tech (doesn't matter for the analysis), and you've classified values into positive and negative. 1e-36 is close enough to zero to make no never mind, and may as well be zero.
In this form where you have (presumably) non-trading day records, like "12" 11 "a" 0 0 -1e-36 you are, in essence, defeating the process of getting a continuous time series, which brings be to the tseries package (https://cran.r-project.org/web/packages/tseries/tseries.pdf ) one of many dealing with a class of problems known as ARIMA.
The key concept is freedom from the calendar; the methods depend only on sequences of events and doesn't need to know about weekends and holiday (trading non-events).
Take a look at getting your data into that type of object and see if that leads you to either solve the problem of the jagged graph or lead you into a new set of questions.
For now, though, I encourage you to focus on your dataset as a trading year where weekends and holidays don't count.
Thanks for the answer, but i didn't really get the idea of the trading year.
An example of the values, that aren't schown in a correct manner, can be seen at timestep 6, where the value "A" changes from positive to negative. This leads to an overlapping of this value at timestep 6 in the plot and it seems like there are positive and negative values at the same time.
ron
December 13, 2018, 8:54am
7
I'm just guessing here - I'd have had a go if your provided data were more easily accessible (eg as output from dput) - but would it work if you interpolate so that you have data points for the points where the both the positive and negative values are zero? That is, for A, a point somewhere between times 6 and 7 something like:
"Time" "Tech" "Values" "positive" "negative"
"1" 0 "a" 15.1537 15.1537 -1e-36
"2" 1 "a" 21.5318 21.5318 -1e-36
"3" 2 "a" 10.4467 10.4467 -1e-36
"4" 3 "a" 21.8219 21.8219 -1e-36
"5" 4 "a" 25.6995 25.6995 -1e-36
"6" 5 "a" 25.9446 25.9446 -1e-36
"7" 6 "a" 21.9551 21.9551 -1e-36
"8" 6.5 "a" 0 0 0
"9" 7 "a" -25.4891 0 -25.4891
... and onwards everywhere that the signs change.
Is there a reason for using -1e-36 rather than 0 for the negative value series?
It will help to deconstruct the data (which will be a lot quicker if you take a look at FAQ: What's a reproducible example (`reprex`) and how do I create one? )
Look at Tech a
There a long periods of no activity, and the area fills in those periods until the next dip (Tech e doesn't seem to be well performing.'
> print(trades_e, n = Inf)
# A tibble: 73 x 3
Time Tech Values
<int> <chr> <dbl>
1 0 e 0
2 1 e 0
3 2 e 0
4 3 e 0
5 4 e 0
6 5 e 0
7 6 e 0
8 7 e 0
9 8 e 0
10 9 e 0
11 10 e -147.
12 11 e -220.
13 12 e -234.
14 13 e -226.
15 14 e -189.
16 15 e -133.
17 16 e -63.5
18 17 e 0
19 18 e 0
20 19 e 0
21 20 e 0
22 21 e 0
23 22 e 0
24 23 e 0
25 24 e 0
26 25 e 0
27 26 e 0
28 27 e 0
29 28 e 0
30 29 e 0
31 30 e 0
32 31 e 0
33 32 e 0
34 33 e 0
35 34 e 0
36 35 e 0
37 36 e 0
38 37 e 0
39 38 e -47.0
40 39 e -36.3
41 40 e 0
42 41 e 0
43 42 e 0
44 43 e 0
45 44 e 0
46 45 e 0
47 46 e 0
48 47 e 0
49 48 e 0
50 49 e 0
51 50 e 0
52 51 e 0
53 52 e 0
54 53 e 0
55 54 e 0
56 55 e 0
57 56 e 0
58 57 e 0
59 58 e 0
60 59 e 0
61 60 e -122.
62 61 e -152.
63 62 e -126.
64 63 e -80.4
65 64 e -33.6
66 65 e 0
67 66 e 0
68 67 e 0
69 68 e 0
70 69 e 0
71 70 e 0
72 71 e 0
73 72 e 0
system
January 4, 2019, 7:30am
9
This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.
