ML nima va qachon kerak

if narx < 500:
    return "tez ketadi"

if narx < 500 and mebel:
    return "tez ketadi"

import pandas as pd
show(pd.DataFrame([
    {"": "An'anaviy dasturlash", "Siz berasiz": "Ma'lumot + QOIDA", "Kompyuter beradi": "Javob"},
    {"": "Machine Learning", "Siz berasiz": "Ma'lumot + JAVOB", "Kompyuter beradi": "QOIDA"},
]))

diagramma(
    {
        "Ma'lumot\n(kvartira xususiyatlari)": (0, 0, 3.2, 1.1, "manba"),
        "QOIDA\n(siz yozasiz: if/else)": (0, 1.6, 3.2, 1.1, "qayta"),
        "Javob\n(tez ketadi / ketmaydi)": (0, 3.2, 3.2, 1.1, "saqlash"),
        "Ma'lumot\n(kvartira xususiyatlari) ": (5.2, 0, 3.2, 1.1, "manba"),
        "Javob\n(tarixdan: ketdi / ketmadi)": (5.2, 1.6, 3.2, 1.1, "manba"),
        "QOIDA\n(model o'zi topadi)": (5.2, 3.2, 3.2, 1.1, "xizmat"),
    },
    [
        ("Ma'lumot\n(kvartira xususiyatlari)", "QOIDA\n(siz yozasiz: if/else)"),
        ("QOIDA\n(siz yozasiz: if/else)", "Javob\n(tez ketadi / ketmaydi)"),
        ("Ma'lumot\n(kvartira xususiyatlari) ", "QOIDA\n(model o'zi topadi)"),
        ("Javob\n(tarixdan: ketdi / ketmadi)", "QOIDA\n(model o'zi topadi)"),
    ],
    nom="an_vs_ml.png",
    izoh="Chapda: an'anaviy dasturlash. O'ngda: machine learning. Qoida va javobning o'rni almashgan.",
)

df = pd.read_csv(DATA / "uy_ijara.csv")
X_misol = df[["xonalar", "maydon_m2", "metro_km", "mebel"]].head(5)
y_misol = df["ijaraga_berildi_30k"].head(5)
print("X — model KO'RADIGAN narsa:")
print(X_misol.to_string(index=False))
print()
print("y — model TOPISHI kerak bo'lgan javob:")
print(y_misol.to_string(index=False))

show(pd.DataFrame([
    {"Holat": "Qoida oddiy va o'zgarmaydi", "ML kerakmi": "YO'Q",
     "Nega": "`if yosh < 18` — buni model bilan qilish ahmoqlik"},
    {"Holat": "Qoida bor, lekin juda murakkab", "ML kerakmi": "HA",
     "Nega": "Yuzni tanish, matnni tushunish — qo'lda yozilmaydi"},
    {"Holat": "Qoida vaqt o'tib o'zgaradi", "ML kerakmi": "HA",
     "Nega": "Bozor narxi, moda, foydalanuvchi xatti-harakati"},
    {"Holat": "Ma'lumot YO'Q yoki juda kam", "ML kerakmi": "YO'Q",
     "Nega": "100 qator bilan model qurilmaydi — avval ma'lumot yig'ing"},
    {"Holat": "Xato juda qimmat, 100% aniqlik kerak", "ML kerakmi": "EHTIYOT",
     "Nega": "ML hech qachon 100% bermaydi. Odam nazorati kerak"},
    {"Holat": "Nega shunday javob berdi — tushuntirish shart", "ML kerakmi": "EHTIYOT",
     "Nega": "Bank krediti, tibbiyot. Oddiy model tanlang (25-bob SHAP)"},
]))

show(pd.DataFrame([
    {"Turi": "Supervised (nazorat ostida)", "Ma'lumot": "X va y — javoblar BOR",
     "Misol": "Kvartira tez ketadimi; narx qancha", "Kitobda": "4-21-bob"},
    {"Turi": "Unsupervised (nazoratsiz)", "Ma'lumot": "Faqat X — javob YO'Q",
     "Misol": "Mijozlarni guruhlarga bo'lish", "Kitobda": "26-28-bob"},
    {"Turi": "Reinforcement (mustahkamlash)", "Ma'lumot": "Harakat va mukofot",
     "Misol": "O'yin, robot boshqaruvi", "Kitobda": "yo'q (alohida soha)"},
]))

print("Bizning faylda IKKITA target bor — ikki xil vazifa:")
print()
print(f"ijaraga_berildi_30k : {sorted(df.ijaraga_berildi_30k.unique().tolist())}  -> {df.ijaraga_berildi_30k.nunique()} xil qiymat")
print("  => CLASSIFICATION (yorliq bashorat qilish)")
print()
print(f"narx_usd            : {df.narx_usd.min()} .. {df.narx_usd.max()}  -> {df.narx_usd.nunique()} xil qiymat")
print("  => REGRESSION (son bashorat qilish)")

show(pd.DataFrame([
    {"": "Classification", "y nima": "Yorliq (toifa)", "Misol": "spam/spam emas, 3 tur gul",
     "Metrika": "accuracy, F1, recall", "Kitobda": "7-9-bob"},
    {"": "Regression", "y nima": "Uzluksiz son", "Misol": "narx, harorat, vaqt",
     "Metrika": "RMSE, MAE, R2", "Kitobda": "14-bob"},
]))

from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier

son = ["xonalar", "maydon_m2", "metro_km", "mebel"]
X, y = df[son], df["ijaraga_berildi_30k"]
X_tr, X_te, y_tr, y_te = train_test_split(X, y, test_size=0.3, random_state=42, stratify=y)

model = DecisionTreeClassifier(max_depth=4, random_state=42)

model.fit(X_tr, y_tr)                       # 1. O'RGAN
bashorat = model.predict(X_te)              # 2. BASHORAT QIL
ball = model.score(X_te, y_te)              # 3. BAHOLA

print(f"fit()     -> model o'rgandi, {len(X_tr)} qatordan")
print(f"predict() -> {len(bashorat)} ta bashorat, masalan: {bashorat[:8].tolist()}")
print(f"score()   -> {ball:.4f}")

from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.neighbors import KNeighborsClassifier

for nom, m in [("DecisionTree", DecisionTreeClassifier(max_depth=4, random_state=42)),
               ("LogisticRegression", LogisticRegression(max_iter=2000)),
               ("RandomForest", RandomForestClassifier(n_estimators=100, random_state=42)),
               ("KNN", KNeighborsClassifier())]:
    m.fit(X_tr, y_tr)
    print(f"  {nom:20s} score = {m.score(X_te, y_te):.4f}")

print(f"{'max_depth':>10} {'train ball':>12} {'test ball':>11} {'farq':>8}")
print("-" * 44)
for d in [2, 4, 8, 15, None]:
    m = DecisionTreeClassifier(max_depth=d, random_state=42).fit(X_tr, y_tr)
    tr_b, te_b = m.score(X_tr, y_tr), m.score(X_te, y_te)
    belgi = "   <-- YODLADI" if tr_b - te_b > 0.15 else ""
    print(f"{str(d):>10} {tr_b:>12.4f} {te_b:>11.4f} {tr_b-te_b:>8.4f}{belgi}")

show(pd.DataFrame([
    {"Holat": "Underfitting", "Train ball": "PAST", "Test ball": "PAST",
     "Sabab": "Model juda oddiy", "Yechim": "Kuchliroq model, ko'proq xususiyat"},
    {"Holat": "Yaxshi", "Train ball": "Yaxshi", "Test ball": "Yaxshi, yaqin",
     "Sabab": "Muvozanat topilgan", "Yechim": "Shu yerda to'xtaysiz"},
    {"Holat": "Overfitting", "Train ball": "JUDA BALAND", "Test ball": "PAST",
     "Sabab": "Model yodlagan", "Yechim": "Soddalashtirish, regularizatsiya, ko'proq ma'lumot"},
]))

show(pd.DataFrame([
    {"Atama": "Feature (xususiyat)", "Ma'nosi": "Kirish ustuni — X", "Bizda": "xonalar, maydon_m2, tuman"},
    {"Atama": "Target (nishon)", "Ma'nosi": "Bashorat qilinadigan ustun — y", "Bizda": "ijaraga_berildi_30k"},
    {"Atama": "Namuna (sample)", "Ma'nosi": "Bitta qator", "Bizda": "Bitta e'lon"},
    {"Atama": "Label (yorliq)", "Ma'nosi": "Classification dagi target qiymati", "Bizda": "0 yoki 1"},
    {"Atama": "Train", "Ma'nosi": "Model o'rganadigan qism", "Bizda": "840 qator"},
    {"Atama": "Test", "Ma'nosi": "Model ko'rmagan qism", "Bizda": "360 qator"},
    {"Atama": "Hyperparameter", "Ma'nosi": "SIZ qo'yadigan sozlama", "Bizda": "max_depth=4"},
    {"Atama": "Parametr (weight)", "Ma'nosi": "Model O'ZI o'rganadigan son", "Bizda": "koeffitsiyentlar"},
]))

show(pd.DataFrame([
    {"Tushuncha": "ML ning mohiyati", "Bitta jumla": "Qoidani siz yozmaysiz, model ma'lumotdan topadi"},
    {"Tushuncha": "Y = f(X)", "Bitta jumla": "Hamma algoritm shuni qiladi, faqat f ni boshqacha izlaydi"},
    {"Tushuncha": "Qachon ML kerak emas", "Bitta jumla": "Qoida oddiy bo'lsa yoki ma'lumot kam bo'lsa"},
    {"Tushuncha": "Classification / regression", "Bitta jumla": "Farq faqat y ning turida — yorliq yoki son"},
    {"Tushuncha": "fit / predict / score", "Bitta jumla": "20 ta algoritm uchun bir xil uch buyruq"},
    {"Tushuncha": "Overfitting", "Bitta jumla": "Train baland, test past — model yodlagan"},
]))