Add plot_probability_matching function for list-based trial data

[Copilot]

The existing plot_probability_matching function expected pandas DataFrame structure with columns like port, rwd, block, etc. After refactoring, parse_arduino_text now returns trial data as a list of dictionaries.

Changes

• New plotting function (src/jdb_to_nwb/plotting/plot_behavior.py)

  • Accepts trial_data (list of dicts) and block_data (list of dicts) from parse_arduino_text
  • Implements rolling window calculations using NumPy instead of pandas
  • Generates 4-panel figure: reward bars for ports A/B/C + rolling choice frequencies with block transitions

• Test coverage (tests/test_convert_behavior.py)

  • Validates plotting with real test data (188 trials, 3 blocks)
  • Verifies file output and figure creation

• Example usage (examples/plot_probability_matching_example.py)

  • Demonstrates integration with parse_arduino_text workflow

Usage

from jdb_to_nwb.convert_behavior import parse_arduino_text
from jdb_to_nwb.plotting.plot_behavior import plot_probability_matching

trial_data, block_data = parse_arduino_text(arduino_text, arduino_timestamps, logger)
fig = plot_probability_matching(trial_data, block_data, fig_dir="./output")

Trial dict structure:

{
    'end_port': 'B',           # Port visited (A/B/C)
    'reward': 1,               # 1 if rewarded, 0 otherwise
    'block': 1,                # Block number
    'trial_within_session': 3  # Trial index
}

Original prompt

def plot_probability_matching(trial_data):
x1 = np.linspace(0,len(trial_data),len(trial_data))
A1 = trial_data.rwd.loc[trial_data.port==0] + 2
yA1 = np.zeros(len(trial_data))
yA1[A1.index.values] = A1
B1 = trial_data.rwd.loc[trial_data.port==1] + 2
yB1 = np.zeros(len(trial_data))
yB1[B1.index.values] = B1
C1 = trial_data.rwd.loc[trial_data.port==2] + 2
yC1 = np.zeros(len(trial_data))
yC1[C1.index.values] = C1

#frequency of each port visit over time
window = 10
#wintype = 'boxcar'

trial_data['ChooseA'] = np.where(trial_data.port==0,1,0)
trial_data['ChooseB'] = np.where(trial_data.port==1,1,0)
trial_data['ChooseC'] = np.where(trial_data.port==2,1,0)

bdt = plt.figure(figsize = (18,12))
plt.suptitle('Within-Session Probability Matching',fontweight='bold',fontsize = 26)
ax4 = plt.subplot2grid((18,1),(3,0),colspan = 1, rowspan =15)
ax4.plot(x1,trial_data.ChooseA.rolling(window,min_periods=1).sum().divide(window),label = 'A',alpha=0.8)
ax4.plot(x1,trial_data.ChooseB.rolling(window,min_periods=1).sum().divide(window),label = 'B',alpha=0.8)
ax4.plot(x1,trial_data.ChooseC.rolling(window,min_periods=1).sum().divide(window),label = 'C',alpha=0.8)
ax4.set_ylabel('Port Visits/trial',fontsize=20,fontweight='bold')
ax4.set_ylim(0,.7)
#ax4.set_xlabel('Time (min)',fontsize=20,fontweight='bold')
ax4.legend(bbox_to_anchor=(0.9, 1.4), loc=2, borderaxespad=0.)
ax1 = plt.subplot2grid((18,1),(0,0),colspan = 1, rowspan =1,sharex=ax4)
ax1.bar(x1,yA1,color = 'blue')
ax1.axis('off')
ax2 = plt.subplot2grid((18,1),(1,0),colspan = 1, rowspan =1,sharex=ax4)
for b in trial_data.block.unique():
    xstart = int((max(trial_data.loc[trial_data.block==b].index)+1))
    ind = max(trial_data.loc[trial_data.block==b].index)
    xmin = int(min(trial_data.loc[trial_data.block==b].index))
    xmid = int(np.mean(trial_data.loc[trial_data.block==b].index))#int(xmin+(xstart-xmin)/2)
    print(xmid)
    if b==1:
        ax4.axvline(x=xstart,color ='r',linestyle='--', label = 'Block Change')
    else:
        ax4.axvline(x=xstart,color ='r',linestyle='--')
    plt.text(xmid-12,8,str(int(trial_data.pA[ind-1]))+': ',fontsize='xx-large',fontweight='bold',color = 'b')
    plt.text(xmid,8,str(int(trial_data.pB[ind-1]))+': ',fontsize='xx-large',fontweight='bold',color = 'orange')
    plt.text(xmid+12,8,str(int(trial_data.pC[ind-1])),fontsize='xx-large',fontweight='bold',color = 'g')
ax4.legend()
ax2.bar(x1,yB1,color='orange')
ax2.axis('off')
ax3 = plt.subplot2grid((18,1),(2,0),colspan = 1, rowspan =1,sharex=ax4)
ax3.bar(x1,yC1,color='g')
ax3.axis('off')
return bdt

I AM Trying to use 'parse_arduino_text' and its 'trial_data', and make probability plot that I used previously with different dataframe structure.
trial_data[i] is looking like this
{'start_time': 41203791.308800004,
'beam_break_start': 41221720.832,
'start_port': 'C',
'end_port': 'B',
'trial_within_block': 3,
'trial_within_session': 3,
'block': 1,
'reward': 1,
'beam_break_end': 41221720.832,
'end_time': 41221720.832}

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[yangsunhwang]

@copilot ---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
Cell In[22], line 1
----> 1 fig = plot_probability_matching(trial_data, block_data, fig_dir="./output")

Cell In[21], line 82, in plot_probability_matching(trial_data, block_data, fig_dir)
79 freq_C = rolling_mean(choose_C, window)
81 # Create figure
---> 82 fig = plt.figure(figsize=(18, 12))
83 plt.suptitle('Within-Session Probability Matching', fontweight='bold', fontsize=26)
85 # Main plot for port visit frequencies

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\pyplot.py:1042, in figure(num, figsize, dpi, facecolor, edgecolor, frameon, FigureClass, clear, **kwargs)
1032 if len(allnums) == max_open_warning >= 1:
1033 _api.warn_external(
1034 f"More than {max_open_warning} figures have been opened. "
1035 f"Figures created through the pyplot interface "
(...) 1039 f"Consider using matplotlib.pyplot.close().",
1040 RuntimeWarning)
-> 1042 manager = new_figure_manager(
1043 num, figsize=figsize, dpi=dpi,
1044 facecolor=facecolor, edgecolor=edgecolor, frameon=frameon,
1045 FigureClass=FigureClass, **kwargs)
1046 fig = manager.canvas.figure
1047 if fig_label:

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\pyplot.py:551, in new_figure_manager(*args, **kwargs)
549 def new_figure_manager(*args, **kwargs):
550 """Create a new figure manager instance."""
--> 551 _warn_if_gui_out_of_main_thread()
552 return _get_backend_mod().new_figure_manager(*args, **kwargs)

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\pyplot.py:528, in _warn_if_gui_out_of_main_thread()
526 def _warn_if_gui_out_of_main_thread() -> None:
527 warn = False
--> 528 canvas_class = cast(type[FigureCanvasBase], _get_backend_mod().FigureCanvas)
529 if canvas_class.required_interactive_framework:
530 if hasattr(threading, 'get_native_id'):
531 # This compares native thread ids because even if Python-level
532 # Thread objects match, the underlying OS thread (which is what
533 # really matters) may be different on Python implementations with
534 # green threads.

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\pyplot.py:369, in _get_backend_mod()
360 """
361 Ensure that a backend is selected and return it.
362
363 This is currently private, but may be made public in the future.
364 """
365 if _backend_mod is None:
366 # Use rcParams._get("backend") to avoid going through the fallback
367 # logic (which will (re)import pyplot and then call switch_backend if
368 # we need to resolve the auto sentinel)
--> 369 switch_backend(rcParams._get("backend"))
370 return cast(type[matplotlib.backend_bases._Backend], _backend_mod)

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\pyplot.py:425, in switch_backend(newbackend)
422 return
423 old_backend = rcParams._get('backend') # get without triggering backend resolution
--> 425 module = backend_registry.load_backend_module(newbackend)
426 canvas_class = module.FigureCanvas
428 required_framework = canvas_class.required_interactive_framework

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\backends\registry.py:317, in BackendRegistry.load_backend_module(self, backend)
303 """
304 Load and return the module containing the specified backend.
305
(...) 314 Module containing backend.
315 """
316 module_name = self._backend_module_name(backend)
--> 317 return importlib.import_module(module_name)

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\importlib_init_.py:90, in import_module(name, package)
88 break
89 level += 1
---> 90 return _bootstrap._gcd_import(name[level:], package, level)

File :1387, in _gcd_import(name, package, level)

File :1360, in find_and_load(name, import)

File :1331, in find_and_load_unlocked(name, import)

File :935, in _load_unlocked(spec)

File :995, in exec_module(self, module)

File :488, in _call_with_frames_removed(f, *args, **kwds)

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\backends\backend_qtagg.py:9
5 import ctypes
7 from matplotlib.transforms import Bbox
----> 9 from .qt_compat import QT_API, QtCore, QtGui
10 from .backend_agg import FigureCanvasAgg
11 from .backend_qt import _BackendQT, FigureCanvasQT

File c:\Users\yangsunhwang\anaconda3\envs\jdb_to_nwb\Lib\site-packages\matplotlib\backends\qt_compat.py:130
128 break
129 else:
--> 130 raise ImportError(
131 "Failed to import any of the following Qt binding modules: {}"
132 .format(", ".join([QT_API for _, QT_API in _candidates]))
133 )
134 else: # We should not get there.
135 raise AssertionError(f"Unexpected QT_API: {QT_API}")

ImportError: Failed to import any of the following Qt binding modules: PyQt6, PySide6, PyQt5, PySide2

not sure we need this qt binding modules?