Ways of visualising larger queues

While vidigi provides a simple way to see the size of larger queues (via the ‘+ x more’ text that appears when the snapshot max is exceeded), this is not particularly visually intuitive for understanding the magnitude of the different queues that are building up.

We could use the synchronised subplot approach from the Additional Synchronised Traces - Orthopaedic Ward - Hospital Efficiency Project example or the gas station example, but when we have multiple steps it could still be challenging to quickly get a sense of where the queues are as your users will have to look across at different parts of the page, and it can be quite a cumbersome thing to add in.

Instead, it would be better to be able to quickly get a sense of queue size directly next to the relevant step. This poses some challenges in vidigi and there are some limitations - but it is possible!

Some of what is shown in this example is planned to be incorporated into the package as an official alternative to the ‘+ x more’ snapshot as part of release 1.1.0. However, seeing under the hood will hopefully help you adapt this approach for your own more complex animations.

We’ll be working with some of the outputs from the first ‘resourceless’ animation example - the mental health appointment booking model.

By the end of that model, we have some queues with as few as 28 people waiting, and some with as high as 462! But because of the way it’s displayed, it’s not easy to get a sense of that.

Let’s kick off with our imports. We’ll be using the separate functions so we can make some changes to the dataframes that get produced along the way.

import pandas as pd
import numpy as np
import re
from vidigi.prep import reshape_for_animations, generate_animation_df
from vidigi.animation import generate_animation
import plotly.io as pio
pio.renderers.default = "notebook"
WARM_UP = 60 * 1
RESULTS_COLLECTION = 90 * 1

We’ve saved the event log produced by the mental health model, so let’s import it here.

event_log_df = pd.read_csv("as-is.csv")

We kick off with a pretty normal call of reshape_for_animations, other than setting the step_snapshot_max to be far higher than what is usually seen, and ensuring it’s so high that the model queue will never reach this level (ensuring we get every entity at every stage).

full_patient_df = reshape_for_animations(
        event_log_df,
        entity_col_name="patient",
        limit_duration=WARM_UP+RESULTS_COLLECTION,
        every_x_time_units=1,
        step_snapshot_max=9999999,
        )

# Remove the warm-up period from the event log
full_patient_df = full_patient_df[full_patient_df["snapshot_time"] >= WARM_UP]

Now, for only the stage we want to display as bars instead of individual entities, we will count the total number of patients queueuing at that stage.

We also need to give each event within this new dataframe a consistent entity ID across all snapshot_times so it will animate correctly.

reshaped_df = full_patient_df.groupby(['snapshot_time', 'event', 'event_type'])['patient'].count().reset_index()

reshaped_df = reshaped_df[(reshaped_df["event"].str.contains("appointment_booked_waiting")) | reshaped_df["event"].str.contains("referred_out")]

event_id_map = {event: idx for idx, event in enumerate(reshaped_df['event'].unique())}

reshaped_df['entity_id'] = reshaped_df['event'].map(event_id_map)
reshaped_df['entity_id'] = reshaped_df['entity_id'].apply(lambda x: f"BAR{x}")
reshaped_df['time'] = reshaped_df['snapshot_time']

reshaped_df = reshaped_df.rename(columns={'patient': 'patient_count'})

reshaped_df.head(40)

	snapshot_time	event	event_type	patient_count	entity_id	time
0	60	appointment_booked_waiting_0	queue	61	BAR0	60
1	60	appointment_booked_waiting_1	queue	9	BAR1	60
2	60	appointment_booked_waiting_10	queue	7	BAR2	60
3	60	appointment_booked_waiting_2	queue	139	BAR3	60
4	60	appointment_booked_waiting_3	queue	52	BAR4	60
5	60	appointment_booked_waiting_4	queue	23	BAR5	60
6	60	appointment_booked_waiting_5	queue	114	BAR6	60
7	60	appointment_booked_waiting_6	queue	61	BAR7	60
8	60	appointment_booked_waiting_7	queue	34	BAR8	60
9	60	appointment_booked_waiting_8	queue	78	BAR9	60
10	60	appointment_booked_waiting_9	queue	212	BAR10	60
22	60	referred_out_0	queue	6	BAR11	60
23	60	referred_out_2	queue	3	BAR12	60
24	60	referred_out_3	queue	1	BAR13	60
25	60	referred_out_4	queue	1	BAR14	60
26	60	referred_out_6	queue	1	BAR15	60
27	60	referred_out_7	queue	1	BAR16	60
28	60	referred_out_8	queue	4	BAR17	60
29	60	referred_out_9	queue	7	BAR18	60
30	61	appointment_booked_waiting_0	queue	69	BAR0	61
31	61	appointment_booked_waiting_1	queue	9	BAR1	61
32	61	appointment_booked_waiting_10	queue	8	BAR2	61
33	61	appointment_booked_waiting_2	queue	148	BAR3	61
34	61	appointment_booked_waiting_3	queue	56	BAR4	61
35	61	appointment_booked_waiting_4	queue	23	BAR5	61
36	61	appointment_booked_waiting_5	queue	117	BAR6	61
37	61	appointment_booked_waiting_6	queue	66	BAR7	61
38	61	appointment_booked_waiting_7	queue	36	BAR8	61
39	61	appointment_booked_waiting_8	queue	81	BAR9	61
40	61	appointment_booked_waiting_9	queue	219	BAR10	61
42	61	referred_out_0	queue	3	BAR11	61
43	61	referred_out_2	queue	3	BAR12	61
44	61	referred_out_7	queue	1	BAR16	61
45	61	referred_out_8	queue	1	BAR17	61
46	61	referred_out_9	queue	5	BAR18	61
47	62	appointment_booked_waiting_0	queue	74	BAR0	62
48	62	appointment_booked_waiting_1	queue	10	BAR1	62
49	62	appointment_booked_waiting_10	queue	8	BAR2	62
50	62	appointment_booked_waiting_2	queue	153	BAR3	62
51	62	appointment_booked_waiting_3	queue	58	BAR4	62

OPTIONAL - but recommend.

Let’s ensure that there’s a count of patients referred out at every point in time. This will ensure we don’t end up with the counts appearing and disappearing across the course of the animation.

# 1a. Extract unique snapshot_times
snapshot_times = reshaped_df['snapshot_time'].unique()

# 1b. Extract unique 'referred_out' event values only
referred_out_events = reshaped_df.loc[reshaped_df['event'].str.startswith('referred_out'), 'event'].unique()

# 2. Create a MultiIndex of all combinations of snapshot_time and referred_out_events
full_index = pd.MultiIndex.from_product(
    [snapshot_times, referred_out_events],
    names=['snapshot_time', 'event']
)

# 3. Set index of the df to ['snapshot_time', 'event'] to facilitate reindexing
df_referred = reshaped_df.set_index(['snapshot_time', 'event'])

# 4. Reindex to full index, filling missing rows with patient_count = 0
df_referred = df_referred.reindex(full_index).reset_index()

# 5. For the new rows where patient_count was filled, fill missing columns like event_type, entity_id, time
df_referred['patient_count'] = df_referred['patient_count'].fillna(0)
df_referred['event_type'] ='queue'
df_referred['time'] = df_referred['snapshot_time']

# Build a dictionary mapping each referred_out event to its corresponding entity_id
event_to_entity = reshaped_df.loc[
    reshaped_df['event'].str.startswith('referred_out'),
    ['event', 'entity_id']
].drop_duplicates().set_index('event')['entity_id'].to_dict()

# Assign using map
df_referred['entity_id'] = df_referred['event'].map(event_to_entity)

# 6. Join this back into our dataframe
final_count_df = pd.concat([
    reshaped_df[reshaped_df['event'].str.contains("appointment_booked_waiting")],
    df_referred
])

We then join this back to the original dataframe, ensuring we first get rid of the entity-level rows for the stage we’ve just done the counts for.

full_patient_df = pd.concat(
    [full_patient_df[(~full_patient_df["event"].str.contains("appointment_booked_waiting")) &
                     (~full_patient_df["event"].str.contains("referred_out"))].rename(columns={"patient": "entity_id"}),
    final_count_df],
    ignore_index=True
)

# If we don't fill the NA values with a float number, we encounter issues later where certain actions can't be performed due to the data type of missing values.
full_patient_df["patient_count"] = full_patient_df["patient_count"].fillna(0.0)

full_patient_df = full_patient_df.sort_values(["snapshot_time", "time", "entity_id"])

Now let’s read in our entity positioning dataframe and carry on with the animation steps.

event_position_df = pd.read_csv("as-is_event_position_df.csv")

# We just use this line to slightly modify the position of one set of our events
event_position_df['x'] = event_position_df.apply(
    lambda row: row['x'] - 120
    if "appointment_booked_waiting" in row['event']
    else row['x'],  # fallback to original value
    axis=1
)

event_position_df

	event	y	x	label	clinic
0	appointment_booked_waiting_0	870.0	505	Booked into<br>clinic 0 🟠	0.0
1	appointment_booked_waiting_1	790.0	505	Booked into<br>clinic 1 🟡	1.0
2	appointment_booked_waiting_2	710.0	505	Booked into<br>clinic 2 🟢	2.0
3	appointment_booked_waiting_3	630.0	505	Booked into<br>clinic 3 🔵	3.0
4	appointment_booked_waiting_4	550.0	505	Booked into<br>clinic 4 🟣	4.0
5	appointment_booked_waiting_5	470.0	505	Booked into<br>clinic 5 🟤	5.0
6	appointment_booked_waiting_6	390.0	505	Booked into<br>clinic 6 ⚫	6.0
7	appointment_booked_waiting_7	310.0	505	Booked into<br>clinic 7 ⚪	7.0
8	appointment_booked_waiting_8	230.0	505	Booked into<br>clinic 8 🔶	8.0
9	appointment_booked_waiting_9	150.0	505	Booked into<br>clinic 9 🔷	9.0
10	appointment_booked_waiting_10	70.0	505	Booked into<br>clinic 10 🟩	10.0
11	have_appointment_0	870.0	850	Attending appointment<br>at clinic 0	NaN
12	have_appointment_1	790.0	850	Attending appointment<br>at clinic 1	NaN
13	have_appointment_2	710.0	850	Attending appointment<br>at clinic 2	NaN
14	have_appointment_3	630.0	850	Attending appointment<br>at clinic 3	NaN
15	have_appointment_4	550.0	850	Attending appointment<br>at clinic 4	NaN
16	have_appointment_5	470.0	850	Attending appointment<br>at clinic 5	NaN
17	have_appointment_6	390.0	850	Attending appointment<br>at clinic 6	NaN
18	have_appointment_7	310.0	850	Attending appointment<br>at clinic 7	NaN
19	have_appointment_8	230.0	850	Attending appointment<br>at clinic 8	NaN
20	have_appointment_9	150.0	850	Attending appointment<br>at clinic 9	NaN
21	have_appointment_10	70.0	850	Attending appointment<br>at clinic 10	NaN
22	referred_out_0	870.0	125	Referred Out From <br>clinic 0	NaN
23	referred_out_1	790.0	125	Referred Out From <br>clinic 1	NaN
24	referred_out_2	710.0	125	Referred Out From <br>clinic 2	NaN
25	referred_out_3	630.0	125	Referred Out From <br>clinic 3	NaN
26	referred_out_4	550.0	125	Referred Out From <br>clinic 4	NaN
27	referred_out_5	470.0	125	Referred Out From <br>clinic 5	NaN
28	referred_out_6	390.0	125	Referred Out From <br>clinic 6	NaN
29	referred_out_7	310.0	125	Referred Out From <br>clinic 7	NaN
30	referred_out_8	230.0	125	Referred Out From <br>clinic 8	NaN
31	referred_out_9	150.0	125	Referred Out From <br>clinic 9	NaN
32	referred_out_10	70.0	125	Referred Out From <br>clinic 10	NaN

full_df_plus_pos = generate_animation_df(
                full_entity_df=full_patient_df,
                entity_col_name="entity_id",
                event_position_df=event_position_df,
                wrap_queues_at=25,
                step_snapshot_max=9999999,
                gap_between_entities=20,
                gap_between_queue_rows=15,
                debug_mode=True
        )

Placement dataframe finished construction at 17:53:58

We’re now going to define a new function that helps us to swap out the counts for an interpretable bar.

def ascii_queue_icon(count, max_count, bar_length=10, filled_char="█", empty_char="░", count_only=False):
    """
    Returns an ASCII progress bar string representing the queue length.

    Args:
        count (int): The current patient count.
        max_count (int): The maximum patient count in the data.
        bar_length (int): Total length of the bar in characters.
        filled_char (str): Character to use for filled segments.
        empty_char (str): Character to use for empty segments.

    Returns:
        str: ASCII progress bar string.
    """
    if count_only:
        if isinstance(count, str):
            return count
        else:
            return f"{count:.0f}"
    else:
        if max_count == 0:
            return empty_char * bar_length  # avoid division by zero

        if not np.isnan(count):
            filled_len = int(round(bar_length * count / max_count))
            bar = filled_char * filled_len + empty_char * (bar_length - filled_len)
            return f"[{bar}] {count:.0f}"

Let’s now add this into our dataframe.

full_df_plus_pos['patient_cumulative'] = full_df_plus_pos.sort_values('snapshot_time').groupby(['event'])['patient_count'].cumsum()
full_df_plus_pos['patient_cumulative_count_display'] = full_df_plus_pos.apply(
    lambda row: f"{row['patient_cumulative']:.0f} (+{row['patient_count']:.0f})",  # fallback to original value
    axis=1
)

# Calculate max patient count only from relevant events
max_count = full_df_plus_pos.loc[
    full_df_plus_pos['event'].str.contains("appointment_booked_waiting"), 'patient_count'
].max()

# Update the icon column conditionally
full_df_plus_pos['icon'] = full_df_plus_pos.apply(
    lambda row: ascii_queue_icon(row['patient_count'], max_count, 20)
    if "appointment_booked_waiting" in row['event']
    else row['icon'],  # fallback to original value
    axis=1
)

full_df_plus_pos['icon'] = full_df_plus_pos.apply(
    lambda row: ascii_queue_icon(row['patient_cumulative_count_display'], max_count, count_only=True)
    if "referred_out" in row['event']
    else row['icon'],  # fallback to original value
    axis=1
)

# Also move the icons over a bit
full_df_plus_pos['x_final'] = full_df_plus_pos.apply(
    lambda row: row['x_final'] - 100
    if "appointment_booked_waiting" in row['event']
    else row['x_final'],  # fallback to original value
    axis=1
)

full_df_plus_pos['x_final'] = full_df_plus_pos.apply(
    lambda row: row['x_final'] - 50
    if "referred" in row['event']
    else row['x_final'],  # fallback to original value
    axis=1
)

From the sample below, we can see some examples of what the icon will now look like.

full_df_plus_pos[full_df_plus_pos["event"].str.contains("appointment_booked_waiting")]

	index	entity_id	pathway	event_type	event	home_clinic	time	booked_clinic	wait	event_original	...	x	label	clinic	x_final	row	y	icon	opacity	patient_cumulative	patient_cumulative_count_display
0	NaN	BAR0	NaN	queue	appointment_booked_waiting_0	NaN	60	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 0 🟠	0.0	405.0	0.0	NaN	[███░░░░░░░░░░░░░░░░░] 61	1.0	61.0	61 (+61)
1	NaN	BAR0	NaN	queue	appointment_booked_waiting_0	NaN	61	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 0 🟠	0.0	405.0	0.0	NaN	[███░░░░░░░░░░░░░░░░░] 69	1.0	130.0	130 (+69)
2	NaN	BAR0	NaN	queue	appointment_booked_waiting_0	NaN	62	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 0 🟠	0.0	405.0	0.0	NaN	[███░░░░░░░░░░░░░░░░░] 74	1.0	204.0	204 (+74)
3	NaN	BAR0	NaN	queue	appointment_booked_waiting_0	NaN	63	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 0 🟠	0.0	405.0	0.0	NaN	[███░░░░░░░░░░░░░░░░░] 73	1.0	277.0	277 (+73)
4	NaN	BAR0	NaN	queue	appointment_booked_waiting_0	NaN	64	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 0 🟠	0.0	405.0	0.0	NaN	[███░░░░░░░░░░░░░░░░░] 73	1.0	350.0	350 (+73)
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
996	NaN	BAR10	NaN	queue	appointment_booked_waiting_9	NaN	146	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 9 🔷	9.0	405.0	0.0	NaN	[███████████████████░] 455	1.0	28610.0	28610 (+455)
997	NaN	BAR10	NaN	queue	appointment_booked_waiting_9	NaN	147	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 9 🔷	9.0	405.0	0.0	NaN	[████████████████████] 461	1.0	29071.0	29071 (+461)
998	NaN	BAR10	NaN	queue	appointment_booked_waiting_9	NaN	148	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 9 🔷	9.0	405.0	0.0	NaN	[████████████████████] 468	1.0	29539.0	29539 (+468)
999	NaN	BAR10	NaN	queue	appointment_booked_waiting_9	NaN	149	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 9 🔷	9.0	405.0	0.0	NaN	[████████████████████] 463	1.0	30002.0	30002 (+463)
1000	NaN	BAR10	NaN	queue	appointment_booked_waiting_9	NaN	150	NaN	NaN	NaN	...	505.0	Booked into<br>clinic 9 🔷	9.0	405.0	0.0	NaN	[████████████████████] 463	1.0	30465.0	30465 (+463)

1001 rows × 24 columns

With this done, we can generate our animation!

generate_animation(
        full_entity_df_plus_pos=full_df_plus_pos,
        event_position_df=event_position_df,
        scenario=None,
        plotly_height=800,
        plotly_width=1000,
        override_x_max=1000,
        override_y_max=1000,
        entity_icon_size=10,
        text_size=10,
        include_play_button=True,
        add_background_image=None,
        display_stage_labels=True,
        time_display_units="d",
        simulation_time_unit="days",
        start_date="2022-06-27",
        setup_mode=False,
        frame_duration=1500, #milliseconds
        frame_transition_duration=1000, #milliseconds
        debug_mode=False
    )

Note that the flow of the patients to their actual appointments is somewhat unintuitive. One simple option is to remove the play button option, which means users can only scrub. This removes any frame interpolation, leading to a more interpretable plot.

generate_animation(
        full_entity_df_plus_pos=full_df_plus_pos,
        event_position_df=event_position_df,
        scenario=None,
        plotly_height=800,
        plotly_width=1000,
        override_x_max=1000,
        override_y_max=1000,
        entity_icon_size=10,
        text_size=10,
        include_play_button=False,   # CHANGED
        add_background_image=None,
        display_stage_labels=True,
        time_display_units="d",
        simulation_time_unit="days",
        start_date="2022-06-27",
        setup_mode=False,
        # frame_duration=1500, #milliseconds
        # frame_transition_duration=1000, #milliseconds
        debug_mode=False
    )

Error changing frame duration
Error changing frame transition duration

We can achieve something similar while not losing the play button by turning the play button back on and setting the frame transition duration to 0.

generate_animation(
        full_entity_df_plus_pos=full_df_plus_pos,
        event_position_df=event_position_df,
        scenario=None,
        plotly_height=800,
        plotly_width=1000,
        override_x_max=1000,
        override_y_max=1000,
        entity_icon_size=10,
        text_size=10,
        include_play_button=True,
        add_background_image=None,
        display_stage_labels=True,
        time_display_units="d",
        simulation_time_unit="days",
        start_date="2022-06-27",
        setup_mode=False,
        frame_duration=1500, #milliseconds
        frame_transition_duration=0, #milliseconds   # CHANGED
        debug_mode=False
    )

Now, in this particular example, it perhaps doesn’t feel like we’re getting a huge amount more information than we would from any old animated bar chart.

So let’s add the final LOS of each individual at the point they attend the clinic, and whether they were a priority patient.

def show_priority_icon(row):
    if "have_appointment" in row["event"]:
        if int(row["pathway"]) == 2:
            return "🚨"
        else:
            return row["icon"]
    else:
        return row["icon"]

def add_los_to_icon(row):
    if "have_appointment" in row["event"]:
        return f'{row["icon"]}<br>{int(row["wait"])}'
    else:
        return row["icon"]

full_df_plus_pos = full_df_plus_pos.assign(
    icon=full_df_plus_pos.apply(show_priority_icon, axis=1)
    )

full_df_plus_pos = full_df_plus_pos.assign(
    icon=full_df_plus_pos.apply(add_los_to_icon, axis=1)
    )

generate_animation(
        full_entity_df_plus_pos=full_df_plus_pos,
        event_position_df=event_position_df,
        scenario=None,
        plotly_height=800,
        plotly_width=1200,
        override_x_max=1000,
        override_y_max=1000,
        entity_icon_size=10,
        text_size=10,
        include_play_button=True,
        add_background_image=None,
        display_stage_labels=True,
        time_display_units="d",
        simulation_time_unit="days",
        start_date="2022-06-27",
        setup_mode=False,
        frame_duration=1000, #milliseconds
        frame_transition_duration=0, #milliseconds   # CHANGED
        debug_mode=False
    )

Values below the individuals show the wait time in days.

Priority patients are identified with a 🚨 symbol. These patients will be moved to the front of the booking queue.

Exploring experimental alternative 1: Replacing ‘x more’ with ascii count bar

def ascii_queue_icon(icon, count, max_count, bar_length=10, filled_char="█", empty_char="░", count_only=False):
    """
    Returns an ASCII progress bar string representing the queue length.

    Args:
        icon (str): The current icon
        count (int): The current patient count.
        max_count (int): The maximum patient count in the data.
        bar_length (int): Total length of the bar in characters.
        filled_char (str): Character to use for filled segments.
        empty_char (str): Character to use for empty segments.

    Returns:
        str: ASCII progress bar string.
    """
    if max_count == 0:
        return empty_char * bar_length  # avoid division by zero

    if not np.isnan(count):
        if count_only:
            return f"{count:.0f}"
        else:
            if "more" in icon:
                filled_len = int(round(bar_length * count / max_count))
                bar = filled_char * filled_len + empty_char * (bar_length - filled_len)
                return f"[{bar}] + {count:.0f} more"
            else:
                return icon


    else:
        return ""

event_log_df = pd.read_csv("as-is.csv")
full_patient_df = reshape_for_animations(
        event_log_df,
        entity_col_name="patient",
        limit_duration=WARM_UP+RESULTS_COLLECTION,
        every_x_time_units=1,
        step_snapshot_max=10,
        )

# Remove the warm-up period from the event log
full_patient_df = full_patient_df[full_patient_df["snapshot_time"] >= WARM_UP]

event_position_df = pd.read_csv("as-is_event_position_df.csv")

event_position_df['x'] = event_position_df.apply(
    lambda row: row['x'] - 70
    if "appointment_booked_waiting" in row['event']
    else row['x'],  # fallback to original value
    axis=1
)

full_df_plus_pos = generate_animation_df(
                full_entity_df=full_patient_df,
                entity_col_name="patient",
                event_position_df=event_position_df,
                wrap_queues_at=10,
                step_snapshot_max=50,
                gap_between_entities=15,
                gap_between_queue_rows=20,
                debug_mode=True
        )

full_df_plus_pos['patient_count'] = full_df_plus_pos.apply(
    lambda row: int(re.search(r'\d+', row['icon'])[0])
    if "more" in row['icon']
    else 0,
    axis=1
)

max_count = max(full_df_plus_pos['patient_count'])

Placement dataframe finished construction at 17:54:06

# Update the icon column conditionally
full_df_plus_pos['icon'] = full_df_plus_pos.apply(
    lambda row: ascii_queue_icon(row['icon'], row['patient_count'], max_count, 10)
    if "appointment_booked_waiting" in row['event']
    else row['icon'],  # fallback to original value
    axis=1
)

# Also move the icon over a bit
full_df_plus_pos['x_final'] = full_df_plus_pos.apply(
    lambda row: row['x_final'] - 100
    if "appointment_booked_waiting" in row['event']
    else row['x_final'],  # fallback to original value
    axis=1
)

generate_animation(
        full_entity_df_plus_pos=full_df_plus_pos,
        event_position_df=event_position_df,
        entity_col_name="patient",
        scenario=None,
        plotly_height=1000,
        plotly_width=1000,
        override_x_max=1000,
        override_y_max=1000,
        entity_icon_size=10,
        text_size=10,
        include_play_button=True,
        add_background_image=None,
        display_stage_labels=True,
        time_display_units="d",
        simulation_time_unit="days",
        start_date="2022-06-27",
        setup_mode=False,
        frame_duration=1500, #milliseconds
        frame_transition_duration=1000, #milliseconds
        debug_mode=False
    )

Let’s again try setting our frame duration to 0.

generate_animation(
        full_entity_df_plus_pos=full_df_plus_pos,
        event_position_df=event_position_df,
        entity_col_name="patient",
        scenario=None,
        plotly_height=1000,
        plotly_width=1000,
        override_x_max=1000,
        override_y_max=1000,
        entity_icon_size=10,
        text_size=10,
        include_play_button=True,
        add_background_image=None,
        display_stage_labels=True,
        time_display_units="d",
        simulation_time_unit="days",
        start_date="2022-06-27",
        setup_mode=False,
        frame_duration=1500, #milliseconds
        frame_transition_duration=30, #milliseconds
        debug_mode=False
    )