Tabulate tracking task data & kinematics

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import os
import json
from datetime import date

import aopy
import aopy.data.bmi3d as bmi3d
import aopy.data.db as db
plt.rcParams['font.size'] = '16'

/home/aolab/miniconda3/envs/katherine-analysis/lib/python3.9/site-packages/pandas/core/computation/expressions.py:21: UserWarning: Pandas requires version '2.8.4' or newer of 'numexpr' (version '2.8.1' currently installed).
  from pandas.core.computation.check import NUMEXPR_INSTALLED
/home/aolab/miniconda3/envs/katherine-analysis/lib/python3.9/site-packages/pandas/core/arrays/masked.py:60: UserWarning: Pandas requires version '1.3.6' or newer of 'bottleneck' (version '1.3.4' currently installed).
  from pandas.core import (
/home/aolab/miniconda3/envs/katherine-analysis/lib/python3.9/site-packages/one/alf/files.py:10: FutureWarning: `one.alf.files` will be removed in version 3.0. Use `one.alf.path` instead.
  warnings.warn(

preproc_dir = '/data/preprocessed/'
db.BMI3D_DBNAME = 'booth' # for in-cage data, use 'tablet'

# load task codes
task_codes = aopy.data.bmi3d.load_bmi3d_task_codes()

# set constants
samplerate = 120

Tabulate (newer) experiment data

Newer tracking task data contains unique event codes and event times for the start and end of ramp periods.

This does not effect the usability of the tabulated dataframe (all columns will be present), simply the behind-the-scenes way the dataframe columns are filled in.

# set subject and range of dates to look up
subject = 'churro'
task_date = (date.fromisoformat('2025-03-03'), date.fromisoformat('2025-03-13'))

# look up sessions (aka entries) of subject within date range
entries = db.lookup_sessions(subject=subject, date=task_date, task_desc='tracking task')
display(entries)

# sort entries into lists of important metadata
subjects, te_ids, te_dates = db.list_entry_details(entries)

[2025-03-03 16:06:16.137089: churro on tracking task, id=19367,
 2025-03-04 16:29:38.509324: churro on tracking task, id=19381,
 2025-03-05 14:59:10.425884: churro on tracking task, id=19390,
 2025-03-05 15:23:08.784877: churro on tracking task, id=19392,
 2025-03-05 15:34:52.672716: churro on tracking task, id=19394,
 2025-03-06 14:21:12.534646: churro on tracking task, id=19401,
 2025-03-06 14:40:30.642826: churro on tracking task, id=19402,
 2025-03-06 15:19:44.195229: churro on tracking task, id=19403,
 2025-03-11 14:46:48.316095: churro on tracking task, id=19423,
 2025-03-11 15:12:35.944435: churro on tracking task, id=19424,
 2025-03-11 15:41:40.775817: churro on tracking task, id=19425]

# tabulate data across entries into a df where rows = trials
metadata = [
    'features', 'scale', 'rotation', 'exp_rotation', 'offset',
    'trajectory_amplitude', 'disturbance_amplitude', 'cursor_radius', 'target_radius',
    'tracking_out_time'
]
df = bmi3d.tabulate_behavior_data_tracking_task(preproc_dir, subjects, te_ids, te_dates, metadata)
df.head()

	subject	te_id	date	event_codes	event_times	reward	penalty	features	scale	rotation	...	tracking_start_time	trajectory_start_time	trajectory_end_time	tracking_end_time	reward_start_time	penalty_start_time	penalty_event	pause_start_time	pause_event	trial_end_time
0	churro	19367	2025-03-03	[16, 2, 64, 239]	[6.02204, 6.457, 6.60116, 7.93624]	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	zyx	...	NaN	NaN	NaN	NaN	NaN	6.60116	64.0	NaN	NaN	7.93624
1	churro	19367	2025-03-03	[16, 2, 81, 97, 81, 97, 81, 80, 96, 80, 96, 80...	[8.11332, 8.52728, 8.93504, 9.3966, 9.72716, 9...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	zyx	...	8.93504	10.48168	30.83520	32.36324	32.36324	NaN	NaN	NaN	NaN	33.00908
2	churro	19367	2025-03-03	[16, 2, 64, 239]	[33.17244, 33.29592, 33.31952, 34.65868]	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	zyx	...	NaN	NaN	NaN	NaN	NaN	33.31952	64.0	NaN	NaN	34.65868
3	churro	19367	2025-03-03	[16, 2, 81, 97, 81, 97, 96, 80, 96, 80, 96, 80...	[34.83156, 35.18776, 35.5954, 36.00936, 36.585...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	zyx	...	35.59540	37.09492	57.57612	59.13304	59.13304	NaN	NaN	NaN	NaN	59.75772
4	churro	19367	2025-03-03	[16, 2, 81, 97, 81, 97, 81, 97, 81, 80, 96, 80...	[59.93952, 59.9478, 60.35448, 60.7454, 61.0116...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	zyx	...	60.35448	61.91224	82.83836	84.39132	84.39132	NaN	NaN	NaN	NaN	85.03296

5 rows × 35 columns

Tabulate (older) experiment data

Older tracking task data does not contain unique ramp events; the start and end of ramp periods are estimated using the sequence generator params.

This does not effect the usability of the tabulated dataframe (all columns will be present), simply the behind-the-scenes way the dataframe columns are filled in.

# set subject and range of dates to look up
subject = 'beignet'
task_date = (date.fromisoformat('2023-06-13'), date.fromisoformat('2023-06-24'))

# look up sessions (aka entries) of subject within date range
entries = db.lookup_sessions(subject=subject, date=task_date, task_desc='tracking task')
display(entries)

# sort entries into lists of important metadata
subjects, te_ids, te_dates = db.list_entry_details(entries)

[2023-06-13 14:29:37.570000: beignet on tracking task, id=9701,
 2023-06-13 14:49:06.959000: beignet on tracking task, id=9702,
 2023-06-13 14:55:33.866000: beignet on tracking task, id=9703,
 2023-06-14 13:39:04.384000: beignet on tracking task, id=9717,
 2023-06-14 13:50:40.366000: beignet on tracking task, id=9718,
 2023-06-14 13:53:52.918000: beignet on tracking task, id=9719,
 2023-06-15 13:08:24.838000: beignet on tracking task, id=9723,
 2023-06-15 13:19:57.469000: beignet on tracking task, id=9724,
 2023-06-15 13:31:18.419000: beignet on tracking task, id=9725,
 2023-06-16 13:53:07.923000: beignet on tracking task, id=9728,
 2023-06-16 14:05:50.244000: beignet on tracking task, id=9729,
 2023-06-16 14:08:57.072000: beignet on tracking task, id=9730,
 2023-06-17 13:40:35.067000: beignet on tracking task, id=9734,
 2023-06-17 13:52:16.907000: beignet on tracking task, id=9735,
 2023-06-17 13:55:26.072000: beignet on tracking task, id=9736,
 2023-06-18 13:34:55.077000: beignet on tracking task, id=9740,
 2023-06-18 13:47:16.040000: beignet on tracking task, id=9741,
 2023-06-18 13:50:46.566000: beignet on tracking task, id=9742,
 2023-06-19 13:36:31.520000: beignet on tracking task, id=9746,
 2023-06-19 13:47:54.395000: beignet on tracking task, id=9747,
 2023-06-19 13:51:05.324000: beignet on tracking task, id=9748,
 2023-06-20 13:19:59.844000: beignet on tracking task, id=9757,
 2023-06-20 13:30:50.048000: beignet on tracking task, id=9758,
 2023-06-20 13:33:48.643000: beignet on tracking task, id=9759,
 2023-06-21 14:34:55.897000: beignet on tracking task, id=9763,
 2023-06-21 14:46:56.112000: beignet on tracking task, id=9764,
 2023-06-21 14:49:53.405000: beignet on tracking task, id=9765,
 2023-06-22 11:34:31.724000: beignet on tracking task, id=9775,
 2023-06-22 11:45:52.258000: beignet on tracking task, id=9776,
 2023-06-22 11:49:13.957000: beignet on tracking task, id=9777,
 2023-06-23 12:54:29.746000: beignet on tracking task, id=9785,
 2023-06-23 13:05:42.598000: beignet on tracking task, id=9786,
 2023-06-23 13:08:54.243000: beignet on tracking task, id=9787]

# tabulate data across entries into a df where rows = trials
metadata = [
    'features', 'scale', 'rotation', 'exp_rotation', 'offset',
    'trajectory_amplitude', 'disturbance_amplitude', 'cursor_radius', 'target_radius',
    'tracking_out_time'
]
df = bmi3d.tabulate_behavior_data_tracking_task(preproc_dir, subjects, te_ids, te_dates, metadata)
df.head()

	subject	te_id	date	event_codes	event_times	reward	penalty	features	scale	rotation	...	tracking_start_time	trajectory_start_time	trajectory_end_time	tracking_end_time	reward_start_time	penalty_start_time	penalty_event	pause_start_time	pause_event	trial_end_time
0	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[6.79096, 7.24928, 7.45828, 11.7332, 12.1748, ...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	7.45828	8.95828	28.97444	30.47444	30.47444	NaN	NaN	NaN	NaN	31.39560
1	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[33.091, 33.09912, 33.30768, 40.55796, 41.2824...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	33.30768	34.80768	54.84080	56.34080	56.34080	NaN	NaN	NaN	NaN	57.26276
2	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[58.95804, 59.224, 59.433, 60.63292, 61.01628,...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	59.43300	60.93300	80.98240	82.48240	82.48240	NaN	NaN	NaN	NaN	83.40760
3	beignet	9701	2023-06-13	[16, 2, 64, 239]	[85.09884, 85.44052, 85.62388, 86.94212]	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	NaN	NaN	NaN	NaN	85.62388	64.0	NaN	NaN	86.94212
4	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[88.64056, 89.09056, 89.29948, 91.6656, 91.916...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	89.29948	90.79948	110.85728	112.35728	112.35728	NaN	NaN	NaN	NaN	113.28256

5 rows × 35 columns

# add additional metadata related to visuomotor perturbations (note: non-perturbation sessions will have NaN values for these params)
param_names = ['pertubation_rotation', 'init_rotation_y', 'final_rotation_y', 'delta_rotation_y', 'trials_per_increment']
param_funcs = [lambda x: x.get_task_param(name) for name in param_names]

db.add_metadata_columns(df, entries, param_names, param_funcs)
df.head()

	subject	te_id	date	event_codes	event_times	reward	penalty	features	scale	rotation	...	penalty_start_time	penalty_event	pause_start_time	pause_event	trial_end_time	pertubation_rotation	init_rotation_y	final_rotation_y	delta_rotation_y	trials_per_increment
0	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[6.79096, 7.24928, 7.45828, 11.7332, 12.1748, ...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	NaN	NaN	NaN	31.39560	6.0	6.0	6.0	6.0	6.0
1	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[33.091, 33.09912, 33.30768, 40.55796, 41.2824...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	NaN	NaN	NaN	57.26276	6.0	6.0	6.0	6.0	6.0
2	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[58.95804, 59.224, 59.433, 60.63292, 61.01628,...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	NaN	NaN	NaN	83.40760	6.0	6.0	6.0	6.0	6.0
3	beignet	9701	2023-06-13	[16, 2, 64, 239]	[85.09884, 85.44052, 85.62388, 86.94212]	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	85.62388	64.0	NaN	NaN	86.94212	6.0	6.0	6.0	6.0	6.0
4	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[88.64056, 89.09056, 89.29948, 91.6656, 91.916...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	NaN	NaN	NaN	113.28256	6.0	6.0	6.0	6.0	6.0

5 rows × 40 columns

# add even more metadata by extracting info from the 'sequence_params' column
df['sequence_params'] = [json.loads(df.iloc[i]['sequence_params']) for i in range(len(df))] # convert 'sequence_params' from string to dict so it's easier to use

time_length = [df.sequence_params[i]['time_length'] for i in range(len(df))]
ramp = [df.sequence_params[i]['ramp'] for i in range(len(df))]
ramp_down = [df.sequence_params[i]['ramp_down'] for i in range(len(df))]

df['time_length'] = time_length
df['ramp'] = ramp
df['ramp_down'] = ramp_down
df.head()

	subject	te_id	date	event_codes	event_times	reward	penalty	features	scale	rotation	...	pause_event	trial_end_time	pertubation_rotation	init_rotation_y	final_rotation_y	delta_rotation_y	trials_per_increment	time_length	ramp	ramp_down
0	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[6.79096, 7.24928, 7.45828, 11.7332, 12.1748, ...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	31.39560	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
1	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[33.091, 33.09912, 33.30768, 40.55796, 41.2824...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	57.26276	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
2	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[58.95804, 59.224, 59.433, 60.63292, 61.01628,...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	83.40760	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
3	beignet	9701	2023-06-13	[16, 2, 64, 239]	[85.09884, 85.44052, 85.62388, 86.94212]	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	86.94212	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
4	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[88.64056, 89.09056, 89.29948, 91.6656, 91.916...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	113.28256	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5

5 rows × 43 columns

Filter tabulated data

# make new df containing only rewarded trials
reward_df = df[df['reward']==True]
reward_df.head()

	subject	te_id	date	event_codes	event_times	reward	penalty	features	scale	rotation	...	pause_event	trial_end_time	pertubation_rotation	init_rotation_y	final_rotation_y	delta_rotation_y	trials_per_increment	time_length	ramp	ramp_down
0	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[6.79096, 7.24928, 7.45828, 11.7332, 12.1748, ...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	31.39560	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
1	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[33.091, 33.09912, 33.30768, 40.55796, 41.2824...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	57.26276	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
2	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[58.95804, 59.224, 59.433, 60.63292, 61.01628,...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	83.40760	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
4	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[88.64056, 89.09056, 89.29948, 91.6656, 91.916...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	113.28256	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
5	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[114.97368, 114.98204, 115.19068, 116.4076, 11...	True	False	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	139.13532	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5

5 rows × 43 columns

# make new df containing only trials that ended in a tracking out penalty
tracking_out_df = df[df.penalty_event == task_codes['OTHER_PENALTY']]
tracking_out_df.head()

	subject	te_id	date	event_codes	event_times	reward	penalty	features	scale	rotation	...	pause_event	trial_end_time	pertubation_rotation	init_rotation_y	final_rotation_y	delta_rotation_y	trials_per_increment	time_length	ramp	ramp_down
10	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[199.44816, 199.73984, 199.94872, 200.18216, 2...	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	207.64756	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
15	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[313.88392, 314.15588, 314.36472, 315.73988, 3...	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	334.70596	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
17	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[339.66408, 339.88904, 340.09812, 341.16664, 3...	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	358.44252	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
21	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 80, 96, 80, 96, 80, 96...	[393.9138, 393.98052, 394.1894, 394.56444, 394...	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	404.47412	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5
22	beignet	9701	2023-06-13	[16, 2, 80, 96, 80, 96, 79, 239]	[406.14708, 409.5304, 409.73932, 411.52264, 41...	False	True	[b'optitrack', b'reward_system', b'saveHDF', b...	1.0	yzx	...	NaN	416.54836	6.0	6.0	6.0	6.0	6.0	20	1.5	1.5

5 rows × 43 columns

Get reward trial kinematics using tabulated data

# specify columns from the df to use for segmenting kinematics
start_event = 'trajectory_start_time'
end_event = 'trajectory_end_time'

# get cursor & target kinematics for rewarded trials
cursor_segments_reward = bmi3d.tabulate_kinematic_data(
    preproc_dir, reward_df['subject'], reward_df['te_id'], reward_df['date'],
    reward_df[start_event], reward_df[end_event],
    datatype='cursor', samplerate=samplerate, upsamplerate=samplerate
)

target_segments_reward = bmi3d.tabulate_kinematic_data(
    preproc_dir, reward_df['subject'], reward_df['te_id'], reward_df['date'],
    reward_df[start_event], reward_df[end_event],
    datatype='reference', samplerate=samplerate, upsamplerate=samplerate
)

# plot kinematics for an example rewarded trial
y_idx = 1 # tracking task signals are 1D, so only have non-zero y-values

for trial_idx in range(5):
    plt.figure(figsize=(10,6))
    plt.title(f'reward trial {trial_idx}', fontsize=18)
    t = np.linspace(0, target_segments_reward[trial_idx].shape[0]/samplerate, target_segments_reward[trial_idx].shape[0])
    plt.plot(t, target_segments_reward[trial_idx][:,y_idx], color='darkorange', label='target')
    plt.plot(t, cursor_segments_reward[trial_idx][:,y_idx], color='blueviolet', label='cursor')
    plt.fill_between(t, target_segments_reward[trial_idx][:,y_idx] - reward_df.target_radius.values[trial_idx],
                        target_segments_reward[trial_idx][:,y_idx] + reward_df.target_radius.values[trial_idx], color='darkorange', alpha=.25)
    plt.fill_between(t, cursor_segments_reward[trial_idx][:,y_idx] - reward_df.cursor_radius.values[trial_idx],
                        cursor_segments_reward[trial_idx][:,y_idx] + reward_df.cursor_radius.values[trial_idx], color='blueviolet', alpha=.25)
    plt.xlabel('time (s)')
    plt.ylabel('position (cm)')
    plt.legend(loc='upper right');

Get penalty trial kinematics using tabulated data

# specify columns from the df to use for segmenting kinematics
start_event = 'trajectory_start_time'
end_event = 'trajectory_end_time' # equivalent to using 'penalty_start_time'

# get cursor & target kinematics for trials that ended in a tracking out penalty
cursor_segments_trackout = bmi3d.tabulate_kinematic_data(
    preproc_dir, tracking_out_df['subject'], tracking_out_df['te_id'], tracking_out_df['date'],
    tracking_out_df[start_event], tracking_out_df[end_event],
    datatype='cursor', samplerate=samplerate, upsamplerate=samplerate
)

target_segments_trackout = bmi3d.tabulate_kinematic_data(
    preproc_dir, tracking_out_df['subject'], tracking_out_df['te_id'], tracking_out_df['date'],
    tracking_out_df[start_event], tracking_out_df[end_event],
    datatype='reference', samplerate=samplerate, upsamplerate=samplerate
)

# plot kinematics for an example penalty trial
y_idx = 1 # tracking task signals are 1D, so only have non-zero y-values

for trial_idx in range(5):
    plt.figure(figsize=(10,6))
    plt.title(f'tracking out trial {trial_idx}', fontsize=18)
    t = np.linspace(0, target_segments_trackout[trial_idx].shape[0]/samplerate, target_segments_trackout[trial_idx].shape[0])
    plt.plot(t, target_segments_trackout[trial_idx][:,y_idx], color='darkorange', label='target')
    plt.plot(t, cursor_segments_trackout[trial_idx][:,y_idx], color='blueviolet', label='cursor')
    plt.fill_between(t, target_segments_trackout[trial_idx][:,y_idx] - tracking_out_df.target_radius.values[trial_idx],
                        target_segments_trackout[trial_idx][:,y_idx] + tracking_out_df.target_radius.values[trial_idx], color='darkorange', alpha=.25)
    plt.fill_between(t, cursor_segments_trackout[trial_idx][:,y_idx] - tracking_out_df.cursor_radius.values[trial_idx],
                        cursor_segments_trackout[trial_idx][:,y_idx] + tracking_out_df.cursor_radius.values[trial_idx], color='blueviolet', alpha=.25)
    plt.xlabel('time (s)')
    plt.ylabel('position (cm)')
    plt.legend(loc='upper right');

Get inter-trial kinematics using tabulated data

# specify columns from the df to use for segmenting kinematics
start_event = 'prev_trial_end_time'
end_event = 'target_on_time'

# get cursor & target kinematics between tracking out penalty and start of next trial
cursor_segments_intertrial = bmi3d.tabulate_kinematic_data(
    preproc_dir, tracking_out_df['subject'], tracking_out_df['te_id'], tracking_out_df['date'],
    tracking_out_df[start_event], tracking_out_df[end_event],
    datatype='cursor', samplerate=samplerate, upsamplerate=samplerate
)

target_segments_intertrial = bmi3d.tabulate_kinematic_data(
    preproc_dir, tracking_out_df['subject'], tracking_out_df['te_id'], tracking_out_df['date'],
    tracking_out_df[start_event], tracking_out_df[end_event],
    datatype='reference', samplerate=samplerate, upsamplerate=samplerate
)

# plot kinematics for an example inter-trial interval
y_idx = 1 # tracking task signals are 1D, so only have non-zero y-values

for trial_idx in range(5):
    plt.figure(figsize=(10,6))
    plt.title(f'inter-trial {trial_idx}', fontsize=18)
    t = np.linspace(0, target_segments_intertrial[trial_idx].shape[0]/samplerate, target_segments_intertrial[trial_idx].shape[0])
    plt.plot(t, target_segments_intertrial[trial_idx][:,y_idx], color='darkorange', label='target')
    plt.plot(t, cursor_segments_intertrial[trial_idx][:,y_idx], color='blueviolet', label='cursor')
    plt.fill_between(t, target_segments_intertrial[trial_idx][:,y_idx] - reward_df.target_radius.values[trial_idx],
                        target_segments_intertrial[trial_idx][:,y_idx] + reward_df.target_radius.values[trial_idx], color='darkorange', alpha=.25)
    plt.fill_between(t, cursor_segments_intertrial[trial_idx][:,y_idx] - reward_df.cursor_radius.values[trial_idx],
                        cursor_segments_intertrial[trial_idx][:,y_idx] + reward_df.cursor_radius.values[trial_idx], color='blueviolet', alpha=.25)
    plt.xlabel('time (s)')
    plt.ylabel('position (cm)')
    plt.legend(loc='upper right');