[Discuss] Add Date column to phase_estimator df

[rebeccadavidsson]

It would be useful if the dataframe also has a 'Date' column, that's now being returned from:

snl.estimate(cs.SIRF, phases=["1st"])
df = snl.phase_estimator(phase="1st").accuracy()

There is a t column, however it's hard to relate this column to the original date.

Temporary Solution

df["Date"] = data.iloc[-len(df):]["Date"].values
df.plot(x="Date", y=["Infected_actual", "Infected_predicted"]);

Also, the y-axis range could be smaller/in range of the data:

Inserting the date would also allow the phase_estimator() function to have multiple phases as input, eventually creating an overview of the accuracy along each phase.

[lisphilar]

Did you try snl.history("Infected")?

# Disable all phases except for 1st and 2nd phases
all_phases = snl.summary().index.tolist()
snl.disable(phases=all_phases)
snl.enable(phases=["1st", "2nd"])
# Parameter estimation
snl.estimate(cs.SIRF)
# Compare actual values and estimated values
snl.history("Infected")

For enhancement, we will add the following features.

• .disable() and .enable() accept phases=None to disable/enable all phases.
• .history() have phases argument to select phases to show.
• .history() accept ylim=(None, None) argument to change y-axis range. (i.e. add **kwargs to control cs.line_plot() function internally.)

If this is a good solution for you, I will create issues later. Do you have time to create pull requests?

[rebeccadavidsson]

Great, thank you!

[lisphilar]

We wiil move forward to #510, #511 and #512.