问题
When I plot my data through rect() (from Bokeh) I get a singular line of horizontal blocks in my visualization. The data prints out correctly and as far as I know formatted correctly (type() verified that they all were lists). Can anyone diagnose this? If the problem is not here then I can append more code.
(If needed: Running Python 2.7.6 on Ubuntu 14.04)
from bokeh.plotting import *
from bokeh.objects import HoverTool, ColumnDataSource
output_notebook()
#All the same color just for testing
colors = [
"#191919", "#191919", "#191919", "#191919", "#191919",
"#191919", "#191919", "#191919", "#191919", "#191919",
"#191919", "#191919", "#191919", "#191919", "#191919",
"#191919", "#191919", "#191919", "#191919", "#191919",
"#191919", "#191919", "#191919", "#191919", "#191919"
]
x_2 = []
for i in trans_dat: x_2.append(i)
y_2 = []
for i in trans_dat.index: y_2.append(i)
colors_2 = []
kwordxstate_2 = []
for y in y_2:
for x in x_2:
kword_state = trans_dat[x][y]
kwordxstate_2.append(kword_state)
colors_2.append(colors[kword_state])
source = ColumnDataSource(
data = dict(
x_2=x_2,
y_2=y_2,
colors_2=colors_2,
kwordxstate_2=kwordxstate_2,
)
)
rect(x_2, y_2, 1,1, source=source,
x_range=x_2, y_range=y_2,
color=colors_2, line_color=None,
tools="resize,hover,previewsave", title="Keywords by state",
plot_width=900, plot_height=400)
grid().grid_line_color = None
axis().axis_line_color = None
axis().major_tick_line_color = None
axis().major_label_text_font_size = "10pt"
axis().major_label_standoff = 0
xaxis().location = "top"
xaxis().major_label_orientation = np.pi/3
show()
回答1:
OK, I'd need to have a complete example with some prototypical trans_dat to dig further. Here are some general comments that might help, though:
x_range and y_range should each be a list of the categories with no duplicates, in the order you want them on the axis.
x and y should be the categorical coordinates for each rect you want to plot. x and y should be the same length.
Immediately it strikes me as odd that you are passing x_2 and y_2 for both the list of categories, and the coordinates. This is usually a mistake.
Let's say you have the following categories:
x-axis:
["US", "Canada"]y-axis:
["Tech", "Agriculture"]
These are what you could pass to x_range and y_range. But if you want a rect for every combination, then you need to pass something like this as x and y:
x:
["US", "US", "Canada", "Canada"]y:
["Tech", Agriculture", "Tech", Agriculture"]
That would result in four rects, one for each pair of categories. If you want to leave some out, that's fine:
x:
["US", "US", "Canada"]y:
["Tech", Agriculture", Agriculture"]
Now there will be no rect for the ("Canada", "Tech") coordinate.
This is analogous to the numerical case: we might have ranges [0,10] and [1,2] for x and y axes. But coordinates are taken from the product of these two ranges, things like (0, 1.5) or (5.5, 2).
Does this make the distinction between the range parameters (which are the list of possible categories) and the coordinate parameters (which are the combinations of categories that you want to plot glyphs at) more clear? Let me know if I can add more info.
来源：https://stackoverflow.com/questions/24179776/issues-with-correlation-graphs-in-bokeh