機械学習　身長体重DB

import sqlite3

dbpath = "./hw.sqlite3"
sql = '''
CREATE TABLE IF NOT EXISTS person (
    id INTEGER PRIMARY KEY,
    height NUMBER,
    weight NUMBER,
    typeNo INTEGER
)
'''
with sqlite3.connect(dbpath) as conn:
    conn.execute(sql)

import sqlite3
import random

dbpath = "./hw.sqlite3"

def insert_db(conn):
    # ダミーで身長と体重と体型データを作る --- (*1)
    height = random.randint(130, 180)
    weight = random.randint(30, 100)
    # 体型データはBMIに基づいて自動生成 --- (*2)
    type_no = 1
    bmi = weight / (height / 100) ** 2
    if bmi < 18.5:
        type_no = 0
    elif bmi < 25:
        type_no = 1
    elif bmi < 30:
        type_no = 2
    elif bmi < 35:
        type_no = 3
    elif bmi < 40:
        type_no = 4
    else:
        type_no = 5
    # SQLと値を指定してDBに値を挿入 --- (*3)
    sql = '''
      INSERT INTO person (height, weight, typeNo) 
      VALUES (?,?,?)
    '''
    values = (height,weight, type_no)
    print(values)
    conn.executemany(sql,[values])

# DBに接続して100件のデータを挿入
with sqlite3.connect(dbpath) as conn:
    # データを100件挿入 --- (*4)
    for i in range(5000):
        insert_db(conn)
    # トータルで挿入した行数を調べる --- (*5)
    c = conn.execute('SELECT count(*) FROM person')
    cnt = c.fetchone()
    print(cnt[0])

import sqlite3

dbpath = "./hw.sqlite3"
select_sql = "SELECT * FROM person"

with sqlite3.connect(dbpath) as conn:
    for row in conn.execute(select_sql):
        print(row)

import keras
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.optimizers import RMSprop

in_size = 2 # 身長と体重の二次元
nb_classes = 6 # 体型を6段階に分ける

# MLPモデル構造を定義
model = Sequential()
model.add(Dense(512, activation='relu', input_shape=(in_size,)))
model.add(Dropout(0.5))
model.add(Dense(nb_classes, activation='softmax'))

# モデルをコンパイル
model.compile(
    loss='categorical_crossentropy',
    optimizer=RMSprop(),
    metrics=['accuracy'])

model.save('hw_model.h5')
print("saved")

from tensorflow.keras.utils import to_categorical
import keras
import numpy as np
import sqlite3
import os
from keras.models import Sequential, load_model
from keras.layers import Dense, Dropout
from keras.optimizers import RMSprop

# データベースから最新の5000件を読み出す
dbpath = "./hw.sqlite3"
select_sql = "SELECT * FROM person ORDER BY id DESC LIMIT 5000"  # 修正

x = []
y = []

with sqlite3.connect(dbpath) as conn:
    for row in conn.execute(select_sql):
        id, height, weight, type_no = row
        # 正規化
        height = height / 200
        weight = weight / 150
        y.append(type_no)
        x.append(np.array([height, weight]))

# モデルを読み込む
model = load_model('hw_model.h5')

# 既に学習データがあれば読み込む
if os.path.exists('hw_weights.weights.h5'):
    model.load_weights('hw_weights.weights.h5')  # ファイル名の変更

nb_classes = 6  # 体型を6段階に分ける
y = to_categorical(y, num_classes=nb_classes)  # 修正

# モデルをコンパイル
model.compile(
    optimizer=RMSprop(),
    loss='categorical_crossentropy',
    metrics=['accuracy']
)

# モデルを訓練
model.fit(np.array(x), y, epochs=100, batch_size=32)

# 結果を保存する
model.save_weights('hw_weights.weights.h5')  # 修正

from keras.models import load_model
import numpy as np

# 学習モデルを読み込む --- (*1)
model = load_model('hw_model.h5')
# 学習済みデータを読み込む --- (*2)
model.load_weights('hw_weights.weights.h5')
# ラベル
LABELS = [
    '低体重(痩せ型)', '普通体重', '肥満(1度)', 
    '肥満(2度)', '肥満(3度)', '肥満(4度)'
]

# テストデータを指定 --- (*3)
height = 160
weight = 50
# 0-1の範囲に収まるようにデータを正規化 --- (*4)
test_x = [height / 200, weight / 150]
# 予測 --- (*5)
pre = model.predict(np.array([test_x]))
idx = pre[0].argmax()
print(LABELS[idx], '/可能性', pre[0][idx])

import os
from keras.models import load_model
import numpy as np
import random
from tensorflow.keras.utils import to_categorical

# 学習モデルを読み込む
model = load_model('hw_model.h5')

# 学習済みデータを読み込む前に存在確認
weights_path = 'hw_weights.weights.h5'
if os.path.exists(weights_path):
    model.load_weights(weights_path)
else:
    print(f"File not found: {weights_path}.")
    # ここで適切な対処を行う
    # 例：モデルの再トレーニング、重みの初期化など

# 正解データを1000件作成
x = []
y = []
for i in range(1000):
    h = random.randint(130, 180)
    w = random.randint(30, 100)
    bmi = w / ((h / 100) ** 2)
    # タイプ番号を決定
    if bmi < 18.5:
        type_no = 0
    elif bmi < 25:
        type_no = 1
    elif bmi < 30:
        type_no = 2
    elif bmi < 35:
        type_no = 3
    elif bmi < 40:
        type_no = 4
    else:
        type_no = 5
    x.append(np.array([h / 200, w / 150]))
    y.append(type_no)

# 形式を変換
x = np.array(x)
y = to_categorical(y, 6)

# 正解率を評価
if model is not None:
    score = model.evaluate(x, y, verbose=1)
    print("正解率=", score[1], "ロス=", score[0])
else:
    print("Model is not loaded correctly.")