Match Naive Bayes to feature types

Understand the idea

Production selects GaussianNB for continuous inputs, BernoulliNB for binary flags and OHE–BernoulliNB for categorical inputs. It does not expose arbitrary mixed inputs or CategoricalNB.

Schematic · Use each production Naive Bayes preparation path.Scroll the diagram horizontally if needed.

Python skill: Uses continuous measurement likelihoods.

Meet the syntax

GaussianNB()
BernoulliNB()
Pipeline([("encode", OneHotEncoder(...)), ("model", BernoulliNB())])

GaussianNB(): Uses continuous measurement likelihoods.
BernoulliNB(): Uses presence and absence of binary indicators.
OneHotEncoder(...): Turns named categories into indicators before the production Bernoulli path; learn its schema inside fitting.

Follow the code

Use the numbered comments to connect each Python block to the workflow above.

from sklearn.naive_bayes import BernoulliNB
flags=['chocolate','fruity','caramel','peanutyalmondy','nougat','crispedricewafer','hard','bar','pluribus']
model=BernoulliNB().fit(train[flags],train.popular)
answer=model.predict(test[flags])

This practice: Read and run the Python. Next: Change · Match Naive Bayes to feature types.

Given data · candy_class

85 observations. One candy product in the survey. The dataframe df is supplied afresh for each Run.

Download source CSV · Source and original dictionary

candy_class · first 8 prepared rows
competitorname	chocolate	fruity	caramel	peanutyalmondy	nougat	crispedricewafer	bar	pluribus	sugarpercent	pricepercent	winpercent	popular
100 Grand	1	0	1	0	0	1	1	0	0.732	0.86	66.9717	50% or above
3 Musketeers	1	0	0	0	1	0	1	0	0.604	0.511	67.6029	50% or above
One dime	0	0	0	0	0	0	0	0	0.011	0.116	32.2611	below 50%
One quarter	0	0	0	0	0	0	0	0	0.011	0.511	46.1165	below 50%
Air Heads	0	1	0	0	0	0	0	0	0.906	0.511	52.3415	50% or above
Almond Joy	1	0	0	1	0	0	1	0	0.465	0.767	50.3475	50% or above
Baby Ruth	1	0	1	1	1	0	1	0	0.604	0.767	56.9145	50% or above
Boston Baked Beans	0	0	0	1	0	0	0	1	0.313	0.511	23.4178	below 50%

Column meanings and units

sugarpercent: sugar percentile. pricepercent: price percentile. winpercent: percentage of survey matchups won. Ingredient, bar and multipack fields are 0/1 flags.

Percentile ranks are neither physical sugar percentages nor currency prices. Ratios of percentiles do not measure economic value. Chocolate and fruit flags can overlap; non-chocolate is not synonymous with fruit. The ML class target is defined by winpercent ≥ 50.

Input schema
Column	Stored type
competitorname	str
chocolate	int64
fruity	int64
caramel	int64
peanutyalmondy	int64
nougat	int64
crispedricewafer	int64
hard	int64
bar	int64
pluribus	int64
sugarpercent	float64
pricepercent	float64
winpercent	float64
popular	str

Supplied setup · available if you need to inspect it

This code runs before your editor on every Run. These are the objects your exercise uses.

from sklearn.model_selection import train_test_split
train,test=train_test_split(df,test_size=.2,random_state=42,stratify=df.popular)

Your task · Follow

Fit BernoulliNB to Candy’s binary flags and predict held-away rows.

Hint 1 — Think

The declared target must stay separate from the presence/absence indicators.

Hint 2 — Tools

BernoulliNB and an explicit flags list.

Hint 3 — Approach

Fit only the nine permitted flags to the training popular labels and predict the same test schema.

Explained solution

from sklearn.naive_bayes import BernoulliNB
flags=['chocolate','fruity','caramel','peanutyalmondy','nougat','crispedricewafer','hard','bar','pluribus']
model=BernoulliNB().fit(train[flags],train.popular)
answer=model.predict(test[flags])

The binary likelihood matches the feature meanings and excludes the target-derived winpercent source.

Helpful prior knowledge: Gaussian Naive Bayes · Encode categories without inventing order These links are guidance, not locks.

Sources and API context

Examples run with this Playground’s scikit-learn 1.4.2 / Pyodide 0.26.4 runtime.