Getting individuals plots from the scorecard ?

[lcrmorin]

After looking trough the documentation It seems that there is no way to plot binning tables from a scorecard.

My 'problem' is that I end up performing the same binning multiple times.

For data exploration I will do something like this:

target = 'loan_status'
features = [c for c in data.columns if c not in target]

for c in features:
    dtype = 'numerical' if pd.api.types.is_numeric_dtype(data[c]) else 'categorical'

    optb = optbinning.OptimalBinning(name=c, dtype=dtype, solver="cp")
    optb.fit(data[c].values, data[target].values.flatten())

    binning_table = optb.binning_table
    binning_table.build()
    binning_table.plot(metric="event_rate")

Then for modelling:

scorecard = Scorecard(binning_process = BinningProcess(features),
                      estimator = LogisticRegression(),
                     )

scorecard.fit(data[features], data[target])

scorecard.table(style="detailed")

The scorecard table provides all the splits. However there is no way to get back to the individual plots easily.

[guillermo-navas-palencia]

Hi @lcrmorin. The scorecard class has the attribute binningprocess, so you can retrieve individual plots from the binning process (see binning process tutorials). Let me know if that works for you.

[lcrmorin]

Indeed, after fitting the scorecard, the following code achieve the same as the data exploration loop presented before.

for c in features:
    scorecard.binning_process_.get_binned_variable(c).binning_table.plot(metric="event_rate")