Exercise 1: penguins, three classes

import pandas as pd

pg = pd.read_csv(DATA / "penguins.csv")
print("shape:", pg.shape)
print()
print(pg.dtypes.to_string())
print()
show(pg.head(6))

print("Classes:")
print(pg.species.value_counts().to_string())
print()
print("ratio:")
print(pg.species.value_counts(normalize=True).round(3).to_string())
print()
missing = pg.isna().sum()
print("Missing values:")
print(missing[missing > 0].to_string())

import pandas as pd
show(pd.DataFrame([
    {"#": 1, "Answer to write down": "How many rows, how many columns"},
    {"#": 2, "Answer to write down": "\"I predict ___ from ___\": one sentence"},
    {"#": 3, "Answer to write down": "Task type: regression / binary / multiclass"},
    {"#": 4, "Answer to write down": "The target ratio and the BASELINE number"},
    {"#": 5, "Answer to write down": "Did you split? Do you need stratify?"},
    {"#": 6, "Answer to write down": "The dummy result"},
    {"#": 7, "Answer to write down": "Which columns you dropped, and WHY"},
    {"#": 8, "Answer to write down": "What you did about missing values and types"},
    {"#": 9, "Answer to write down": "Your first honest result"},
    {"#": 10, "Answer to write down": "Which metric you chose, and WHY"},
    {"#": 11, "Answer to write down": "Which group the mistakes cluster in"},
    {"#": 12, "Answer to write down": "Which ONE change you made, and the result"},
]))

from sklearn.model_selection import train_test_split, cross_val_score
from sklearn.pipeline import Pipeline
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.impute import SimpleImputer
from sklearn.linear_model import LogisticRegression
from sklearn.dummy import DummyClassifier
from sklearn.metrics import accuracy_score, f1_score, confusion_matrix, classification_report

num = ["bill_length_mm", "bill_depth_mm", "flipper_length_mm", "body_mass_g"]
cat = ["island", "sex"]
X, y = pg[num + cat], pg["species"]
X_tr, X_te, y_tr, y_te = train_test_split(X, y, test_size=0.3, random_state=42, stratify=y)

baseline = accuracy_score(y_te, DummyClassifier(strategy="most_frequent").fit(X_tr[num], y_tr).predict(X_te[num]))

pipe = Pipeline([
    ("prep", ColumnTransformer([
        ("num", Pipeline([("impute", SimpleImputer(strategy="median")),
                          ("scale", StandardScaler())]), num),
        ("cat", Pipeline([("impute", SimpleImputer(strategy="most_frequent")),
                          ("encode", OneHotEncoder(handle_unknown="ignore"))]), cat),
    ])),
    ("model", LogisticRegression(max_iter=3000)),
]).fit(X_tr, y_tr)

b = pipe.predict(X_te)
print(f"BASELINE (most_frequent): {baseline:.4f}")
print(f"Model accuracy          : {accuracy_score(y_te, b):.4f}")
print(f"Model macro F1          : {f1_score(y_te, b, average='macro'):.4f}")
print(f"Model weighted F1       : {f1_score(y_te, b, average='weighted'):.4f}")
cv = cross_val_score(pipe, X, y, cv=5, scoring="f1_macro")
print(f"CV macro F1             : {cv.mean():.4f} ±{cv.std():.4f}")

print("The 3x3 confusion matrix:")
classes = sorted(y.unique())
cm = confusion_matrix(y_te, b, labels=classes)
print(pd.DataFrame(cm, index=[f"truth: {s}" for s in classes],
                   columns=[f"predicted: {s}" for s in classes]).to_string())
print()
print(classification_report(y_te, b, digits=4))

print("The island column against the target:")
print(pd.crosstab(pg.island, pg.species).to_string())