Adding computation and plotting of trial by trial metrics

[alexpiet]

Updated plot_session_scroller to plot metrics

• users can specify which metrics to plot by passing in a list of metrics to plot
• metrics are plotted with or without FIP data
• metrics are randomly colored, we might want to hardcode a color mapping later
• added a marker to plot whether the port is baited

Trial by trial metrics:

• response_rate, fraction of trials with a response
• gocue_reward_rate, fraction of trials with a reward
• response_reward_rate, fraction of trials with a reward computed only on trials with a response
• choose_right_rate, fraction of trials where mouse chose right computed only on trials with a response
• pL, generative reward probability on the left, just added plotting
• pR, generative reward probability on the right, just added plotting

Future work:

• There are many more metrics to add, but I wanted to merge this in so we have a stable ecosystem for metrics.
• parameters for computing metrics (window duration, minimum number of events, whether to "center" the window) might want to be metric specific

DEMO

nwb.df_trials = compute_all_trial_metrics(nwb)
plot_session_scroller(nwb, plot_bouts=True, metrics=['response_rate','choose_right_rate'])

Another example with baiting and generative reward probabilities

nwb.df_trials = compute_all_trial_metrics(nwb)
nwb.fip_df = create_fib_df(nwb)
plot_session_scroller(nwb, plot_bouts=True, metrics=['pR','pL'])

[hanhou]

• See this issue add instantaneous stats for each session foraging-behavior-browser#2

[alexpiet]

@hanhou @rachelstephlee @nkeesey @ZhixiaoSu This is ready for review. I'd like to merge it in quickly. It would be good to agree on names for metrics

[hanhou]

I have more trial-based metrics here that can be migrated to this function.

Most of my metrics are not binary variables, so they may not need smoothing.

[alexpiet]

Tracking them in this issue: #27

[hanhou]

Now the trial metrics are plotted in time-based plot. Do you think it makes sense to also add trial metrics to the trial-based session plot as well?

Also, it would be great to have a plotly version of all these plots, esp time-based plots with FIP. Then it will be very easy to visualize in Streamlit or Jupyter notebook

[hanhou]

I'll go ahead and approve this for now.

[alexpiet]

• See this issue add instantaneous stats for each session foraging-behavior-browser#2

Tracking them in this consolidated issue: #27

[alexpiet]

trial-based session plot

Yes, made a separate issue here: #28