Plotly px.Timeline y marks do not adjust when using facet_row

Question

With a simple dataFrame that look like this

The following code

fig = px.timeline(df, 
              x_start="Target Date Start", 
              x_end="Target Date End", 
              y="Initiative", 
              color="Status Bin", 
              facet_row="Project Key",
              template="plotly_white") 

Generates a graph that does not adjust based on facet row. I would expect only initiatives associated with a given project key to show in the y marks, but instead all initiatives are shown:

Note that it shows only the relevant bars but it shows all y marks as opposed to filter for relevant initiatives and also it keeps the size the same across facet_rows whereas I would expect their size to be proportional to the number of initiatives in their resp. group

EDIT: r-beginners suggested fix

Answer 1

If your goal is to optimize the y-axis for each individual subplot, rather than to make the y-axis values common to all subplots, then a graph object can handle this, but it will break the graph visualized in the well-organized x and y axes.

import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots

row_cnt = len(df['Project Key'].unique())
fig = make_subplots(rows=row_cnt, cols=1)

for idx,k in enumerate(df['Project Key'].unique()):
    dff = df[df['Project Key'] == k]
    print(dff)
    tl = px.timeline(dff,
                      x_start="Target Date Start",
                      x_end="Target Date End",
                      y="Initiative",
                      color="Status Bin")
    for t in range(len(tl.data)):
        fig.add_trace(tl.data[t], row=idx+1, col=1)
        fig.update_traces(width=0.1)
        fig.update_traces(legendgroup=k, legendgrouptitle_text=f'Project Key:{k}', row=idx+1, col=1)
        fig.update_xaxes(type='date', range=[df['Target Date Start'].min(),df['Target Date End'].max()]) 
    
fig.update_layout(height=1200, width=1000, template="plotly_white")

fig.show()