How to add a OLS trendline to a plotly scatter plot graph object that uses updatemenus to display subsets of data?
I am trying to make a plotly scatter plot with a OLS trendline and provide a dropdown menu what let's the user pick from a different of X's and Y's. The plot almost works. Where the first figure produced by the code has the data and the trendline, but when you choose an option from the drop down the trendline instead connects every done with another dot, seeminly in the order the coordinated appear in the data frame. `
The code:
import plotly.graph_objects as go
import plotly.express as px
#making a figure
fig = go.Figure()
x1 = df['crttotal']
x2 = df['nfcc_mean']
x3 = df['bficonmean']
x4 = df['bfiopmean']
y1 = df['avg_misperception_score']
y2 = df['avg_ambiguous_score']
fig = px.scatter(df, x=x2, y=y1, trendline="ols", trendline_scope="overall")
#making the dropdown
fig.update_layout(
updatemenus=[
go.layout.Updatemenu(
type="dropdown",
buttons=list([
dict(label="NFCC vs Misperception",
method="update",
args=[{"x": [x2], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "NFCC vs Misperception"}]),
dict(label="CRT vs Misperception",
method="update",
args=[{"x": [x1], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "CRT vs Misperception"}]),
dict(label="bficonmean vs Misperception",
method="update",
args=[{"x": [x3], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "bficonmean vs Misperception"}]),
dict(label="bfiopmean vs Misperception",
method="update",
args=[{"x": [x4], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "bfiopmean vs Misperception"}]),
dict(label="CRT vs Ambiguity",
method="update",
args=[{"x": [x1], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "CRT vs Ambiguity"}]),
dict(label="NFCC vs Ambiguity",
method="update",
args=[{"x": [x2], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "NFCC vs Ambiguity"}]),
dict(label="bficonmean vs Ambiguity",
method="update",
args=[{"x": [x3], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "bficonmean vs Ambiguity"}]),
dict(label="bfiopmean vs Ambiguity",
method="update",
args=[{"x": [x4], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "bfiopmean vs Ambiguity"}])
])
)
]
)
#set the title
fig.update_layout(title="Dropdown")
fig.show()
the Data
crttotal nfcc_mean bficonmean bfiopmean avg_misperception_score \
0 3 2.87 3.875 3.000 -0.062
1 0 3.53 3.625 3.125 -0.235
2 0 3.80 4.000 3.000 0.077
3 0 3.73 3.750 3.500 0.067
4 2 3.87 3.125 3.000 0.368
5 0 3.47 2.750 3.500 -0.200
6 0 4.33 3.625 3.625 -0.200
7 0 4.13 3.250 3.125 -0.500
8 0 4.73 3.250 3.250 -0.643
9 3 5.20 3.750 2.750 0.000
avg_ambiguous_score
0 2.60
1 2.10
2 3.35
3 2.55
4 2.90
5 2.80
6 2.85
7 3.30
8 3.15
9 2.70
What I was expecting (and what initially shows when i run the above code)enter image description here
What I get when I change select an option from my dropdown: enter image description here
** EDIT: **
"AS DISCUSSED IN COMMENT" your solution worked, but the trenline info box always reads with the same equation: avg_misperception = -.004*crttotal + .073
How do I get it so that the info box also updates? [![Trendline info box does not match graph variables][1]][1] [![Trendline info box does not match graph variables][2]][2]
Seriously, you're amazing. TY so much for providing such a clear answer. [1]: https://i.sstatic.net/zPxBL.png [2]: https://i.sstatic.net/fy7m6.png
EDIT: the same is the case for the data point labels. When you hover over a data their values are labeled as "crttotal" and "avg_misperception_score." I assume I need to update something in the Layout Attribute - https://plotly.com/python/figure-structure/#the-toplevel-layout-attribute? trying to figure it out on my own as well - just new to plotly.
Short answer
Add more data in the lists for "x" and "y" in {"x": [x2, <regression x>], "y": [y1, <regression y>], ...} to display a trendline. Of course, you'll need to calculate those values somehow. The suggestion below shows one possible way to do exactly that.
The details
The reason why you're not seeing trendlines after clicking the button options is that there is no definition and/or data for a trendline in, for example:
args=[{"x": [x2], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "NFCC vs Misperception"}]
And this is not very apparent, but you could easily add data for more traces in the layout that the button-click creates by adding more data in the lists associated with "x"and "y" like this:
args=[{"x": [x2, x21], "y": [y1, y11], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "NFCC vs Misperception"}]
How you choose to create that data is up to you. But sticking to Plotly Express, you can create and retrieve data for an OLS regression like this:
px.scatter(x=x2, y=y1, trendline='ols').data[1].x
And:
px.scatter(x=x2, y=y1, trendline='ols').data[1].y
Using that, the args for your first button would look like this:
args=[{"x": [x2, px.scatter(x=x2, y=y1, trendline='ols').data[1].x], "y": [y1, px.scatter(x=x2, y=y1, trendline='ols').data[1].y], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "NFCC vs Misperception"}]
Here's a complete setup for your two first buttons with some random data:
Plot1:
Plot2:
If you would share your data in your code snippet with the output from df.to_dict() I might find the time to write up a complete solution.
Code:
import plotly.graph_objects as go
import plotly.express as px
import numpy as np
import pandas as pd
# making a figure
df = pd.DataFrame({'crttotal': np.random.random(8),
'nfcc_mean': np.random.random(8),
'bficonmean': np.random.random(8),
'bfiopmean': np.random.random(8),
'avg_misperception_score': np.random.random(8),
'avg_ambiguous_score': np.random.random(8)})
fig = go.Figure()
x1 = df['crttotal']
x2 = df['nfcc_mean']
x3 = df['bficonmean']
x4 = df['bfiopmean']
y1 = df['avg_misperception_score']
y2 = df['avg_ambiguous_score']
fig = px.scatter(df, x=x2, y=y1, trendline="ols", trendline_scope="overall")
fig.show()
# making the dropdown
fig.update_layout(
updatemenus=[
go.layout.Updatemenu(
type="dropdown",
buttons=list([
dict(label="NFCC vs Misperception",
method="update",
args=[{"x": [x2, px.scatter(x=x2, y=y1, trendline='ols').data[1].x], "y": [y1, px.scatter(x=x2, y=y1, trendline='ols').data[1].y], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "NFCC vs Misperception"}]),
dict(label="CRT vs Misperception",
method="update",
args=[{"x": [x1, px.scatter(x=x1, y=y1, trendline='ols').data[1].x], "y": [y1, px.scatter(x=x1, y=y1, trendline='ols').data[1].y], "trendline":["ols"], "trendline_scope":["overall"]},
{"title": "CRT vs Misperception"}]),
# dict(label="bficonmean vs Misperception",
# method="update",
# args=[{"x": [x3], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
# {"title": "bficonmean vs Misperception"}]),
# dict(label="bfiopmean vs Misperception",
# method="update",
# args=[{"x": [x4], "y": [y1], "trendline":["ols"], "trendline_scope":["overall"]},
# {"title": "bfiopmean vs Misperception"}]),
# dict(label="CRT vs Ambiguity",
# method="update",
# args=[{"x": [x1], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
# {"title": "CRT vs Ambiguity"}]),
# dict(label="NFCC vs Ambiguity",
# method="update",
# args=[{"x": [x2], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
# {"title": "NFCC vs Ambiguity"}]),
# dict(label="bficonmean vs Ambiguity",
# method="update",
# args=[{"x": [x3], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
# {"title": "bficonmean vs Ambiguity"}]),
# dict(label="bfiopmean vs Ambiguity",
# method="update",
# args=[{"x": [x4], "y": [y2], "trendline":["ols"], "trendline_scope":["overall"]},
# {"title": "bfiopmean vs Ambiguity"}])
])
)
]
)
# set the title
fig.update_layout(title="Dropdown")
# fig.show()
- Collatz's hypothesis || want to reuse code instead of repeating
- why my directory copy endless when run copytree command of shutil?
- How can I reconnect to the browser opened by webdriver with selenium?
- Create and use WAV file as an object Python
- How to create a rank or an index column based on more than one column?
- Polars pairwise sum of array column
- Pandas not recognising older dates (pre-1600)
- How to make sense of this data extraction and convert it to pandas dataframe?
- How does Python's unittest library match patterns passed via the -p parameter?
- Python github composite action returns name not defined error
- How to run pytest-cov with pytest in VS Code?
- How do I append a string to a Path?
- What does a bare asterisk do in a parameter list? What are "keyword-only" parameters?
- Amazon textextract I can't find trp module
- Pydantic TypeError: validate() takes 2 positional arguments but 3 were given
- Attempted relative import with no known parent package
- How To Subscribe To Websocket API Channel Using Python?
- rdkit.Chem.rdForceFieldHelpers.UFFOptimizeMolecule(int) did not match C++ signature
- Connecting to SQL Server via SQL magics in Jupyter Lab
- Using httplib2 in python 3 properly? (Timeout problems)
- Incorrect calculation in the list processing logic based on dependencies between the elements of these lists
- Possible import error for pandas and matplotlib when calling Python3 via PHP shell_exec
- Attributes on Python flag enums
- Upload content from GET request to Cloud Storage
- mpmath : cannot create mpf
- DLL load failed when importing PyQt5
- How can I replace a substring in a Python pathlib.Path?
- ModuleNotFoundError: No module named 'polls'
- pyvenv-3.4 returned non-zero exit status 1
- stripe.error.SignatureVerificationError: No signatures found matching the expected signature for payload" - Python FastAPI