יסודות בינה מלאכותית : 10-RB27 – מבוא לרשת ניירונים

מחבר:admin
פורסם:דצמבר 3, 2024
קטגוריה:רובוטרוניקס כללי
תגובות:אין תגובות

יסודות בינה מלאכותית : 10-RB27 – מבוא לרשת ניירונים

תרגול כיתה יחד : התקנה של סביבית עבודה – לבינה מלאכותית עבור אנקודנה שלב אחר שלב
ניפתח את :
ונפעיל את התוכנית

conda create -n face_env python=3.8

conda activate face_env

pip install opencv-python

pip install numpy==1.21

for face detection :

pip install dlib

pip install face_recognition

pip install matplotlib

conda install jupyter notebook -y

jupyter notebook –version

2. הרצה של בינה מלאכותית מאוד פשוטה – לזיהוי פנים הורדת תמונות – תרגול כיתה

3. רשות ניירונים

4. ניתוח קובץ אקסל ושימוש ברשת ניירונים

4.1 שים לב שאנחנו ב BASE באנקודנה

4.2 התקנה סיפריות אם צריך

pip install pandas
pip install numpy
pip install tensorflow
pip install scikit-learn
pip install matplotlib

 tqdm (optional, for a progress bar in Jupyter Notebooks) #
pip install tqdm

pip install pandas

pip install numpy

pip install tensorflow

pip install scikit-learn

pip install matplotlib

tqdm (optional, for a progress bar in Jupyter Notebooks) #

pip install tqdm

4.2.1 נבדוק שהכל הותקן

import pandas as pd
import numpy as np
import tensorflow as tf
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt

print("All packages imported successfully!")

import pandas as pd

import numpy as np

import tensorflow as tf

from sklearn.model_selection import train_test_split

from sklearn.preprocessing import StandardScaler

import matplotlib.pyplot as plt

print("All packages imported successfully!")

pip install pandas seaborn matplotlib

1	pip install pandas seaborn matplotlib

4.3 נוריד את הקובץ עבוד בינה מלאכותית מסוג ANN

USA Housing Dataset1

4.4 נריץ את הקוד באנקודנה Jupiter notebook

import os
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Check if the file exists
file_path = r'd:\temp\USA Housing Dataset1.csv'
if os.path.exists(file_path):
    print("File found.")
else:
    raise FileNotFoundError("File not found. Check the path.")

# Load the dataset
data = pd.read_csv(file_path)

# Display the first few rows to understand the structure
print("Dataset Preview:")
print(data.head())

# Remove rows where 'price-usd' is zero or negative
data = data[data['price-usd'] > 0]

# Display basic statistics
print("\nDataset Statistics:")
print(data.describe())

# Select numeric columns for correlation analysis
numeric_features = data.select_dtypes(include=['float64', 'int64']).columns

# Correlation matrix
correlation_matrix = data[numeric_features].corr()

# Plot heatmap of correlation matrix
plt.figure(figsize=(10, 8))
sns.heatmap(correlation_matrix, annot=True, fmt='.2f', cmap='coolwarm', cbar=True)
plt.title('Correlation Heatmap')
plt.show()

# Analyze correlation with house price
price_correlation = correlation_matrix['price-usd'].sort_values(ascending=False)
print("\nCorrelation with 'price-usd':")
print(price_correlation)

# Scatter plots for top correlated features
top_features = price_correlation.index[1:4]  # Select top 3 features excluding 'price-usd'

for feature in top_features:
    plt.figure(figsize=(6, 4))
    sns.scatterplot(data=data, x=feature, y='price-usd')
    plt.title(f"Price vs {feature}")
    plt.xlabel(feature)
    plt.ylabel('Price (USD)')
    plt.show()

# Save the correlation matrix to a CSV file
correlation_output_path = r'd:\temp\correlation_matrix.csv'
correlation_matrix.to_csv(correlation_output_path)
print(f"Correlation matrix saved to: {correlation_output_path}")

import os

import pandas as pd

import seaborn as sns

import matplotlib.pyplot as plt

# Check if the file exists

file_path = r'd:\temp\USA Housing Dataset1.csv'

if os.path.exists(file_path):

print("File found.")

else:

raise FileNotFoundError("File not found. Check the path.")

# Load the dataset

data = pd.read_csv(file_path)

# Display the first few rows to understand the structure

print("Dataset Preview:")

print(data.head())

# Remove rows where 'price-usd' is zero or negative

data = data[data['price-usd'] > 0]

# Display basic statistics

print("\nDataset Statistics:")

print(data.describe())

# Select numeric columns for correlation analysis

numeric_features = data.select_dtypes(include=['float64', 'int64']).columns

# Correlation matrix

correlation_matrix = data[numeric_features].corr()

# Plot heatmap of correlation matrix

plt.figure(figsize=(10, 8))

sns.heatmap(correlation_matrix, annot=True, fmt='.2f', cmap='coolwarm', cbar=True)

plt.title('Correlation Heatmap')

plt.show()

# Analyze correlation with house price

price_correlation = correlation_matrix['price-usd'].sort_values(ascending=False)

print("\nCorrelation with 'price-usd':")

print(price_correlation)

# Scatter plots for top correlated features

top_features = price_correlation.index[1:4] # Select top 3 features excluding 'price-usd'

for feature in top_features:

plt.figure(figsize=(6, 4))

sns.scatterplot(data=data, x=feature, y='price-usd')

plt.title(f"Price vs {feature}")

plt.xlabel(feature)

plt.ylabel('Price (USD)')

plt.show()

# Save the correlation matrix to a CSV file

correlation_output_path = r'd:\temp\correlation_matrix.csv'

correlation_matrix.to_csv(correlation_output_path)

print(f"Correlation matrix saved to: {correlation_output_path}")

4.3 רשת ניירונים וניתוח קובץ אקבל בקוד ( נשווה מול CHATGPT מול קוד בהמשך )

import pandas as pd
import numpy as np
import tensorflow as tf
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt
import os

# Load the dataset
file_path = r'd:\temp\USA Housing Dataset1.csv'
if os.path.exists(file_path):
    print("File found.")
else:
    print("File not found. Check the path.")

data = pd.read_csv(file_path)
print("File loaded into memory.")

# Remove rows where 'price-usd' is zero
data = data[data['price-usd'] > 0]

# Normalize target (price)
data['price-usd'] = data['price-usd'] / 1e6  # Scale target to millions of dollars

# Prepare features and target
X = data[['bedrooms', 'bathrooms', 'sqft_living', 'sqft_lot', 'floors',
          'waterfront', 'view', 'condition', 'sqft_above', 'sqft_basement', 'yr_built']].values
y = data['price-usd'].values

# Split the data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Standardize the features
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)

# Define the ANN model
input_size = X_train.shape[1]
hidden_size = 32
output_size = 1

# Define the model with Input layer
model = tf.keras.Sequential([
    tf.keras.layers.Input(shape=(input_size,)),  # Explicit input shape
    tf.keras.layers.Dense(hidden_size, activation='relu'),
    tf.keras.layers.Dense(hidden_size, activation='relu'),
    tf.keras.layers.Dense(output_size)
])

# Compile the model
model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.0001),
              loss='mse')

# Train the model
history = model.fit(X_train, y_train, epochs=1000, validation_data=(X_test, y_test), verbose=1)

# Plot training and validation loss
plt.figure(figsize=(10, 6))
plt.plot(history.history['loss'], label='Training Loss', marker='o')
plt.plot(history.history['val_loss'], label='Validation Loss', marker='x')
plt.title('Training and Validation Loss History')
plt.xlabel('Epoch')
plt.ylabel('Loss')
plt.legend()
plt.grid(True)
plt.show()

# Evaluate the model on the test dataset
y_pred_test = model.predict(X_test).flatten()

# Calculate accuracy for each prediction
accuracy = 100 * (1 - np.abs(y_pred_test - y_test) / y_test)

# Handle cases where actual price is zero (to avoid division by zero)
accuracy = np.where(y_test == 0, 0, accuracy)

# Create a DataFrame for predictions, actual prices, and accuracy
evaluation_df = pd.DataFrame({
    'Predicted Price (Million $)': y_pred_test,
    'Actual Price (Million $)': y_test,
    'Accuracy (%)': accuracy
})

# Display first 10 rows of the table
print("\nEvaluation: Predicted vs Actual Prices with Accuracy")
print(evaluation_df.head(10))

# Save the evaluation table to a CSV file
evaluation_df.to_csv(r'd:\temp\evaluation_results_with_accuracy.csv', index=False)

# Plot Predicted vs Actual Prices
plt.figure(figsize=(10, 6))
plt.scatter(evaluation_df['Actual Price (Million $)'], evaluation_df['Predicted Price (Million $)'], alpha=0.7)
plt.xlabel('Actual Price (Million $)')
plt.ylabel('Predicted Price (Million $)')
plt.title('Predicted vs Actual Prices')
plt.plot([evaluation_df['Actual Price (Million $)'].min(), evaluation_df['Actual Price (Million $)'].max()],
         [evaluation_df['Actual Price (Million $)'].min(), evaluation_df['Actual Price (Million $)'].max()],
         color='red', linestyle='--', label='Perfect Prediction Line')
plt.legend()
plt.grid(True)
plt.show()

100

101

102

103

import pandas as pd

import numpy as np

import tensorflow as tf

from sklearn.model_selection import train_test_split

from sklearn.preprocessing import StandardScaler

import matplotlib.pyplot as plt

import os

# Load the dataset

file_path = r'd:\temp\USA Housing Dataset1.csv'

if os.path.exists(file_path):

print("File found.")

else:

print("File not found. Check the path.")

data = pd.read_csv(file_path)

print("File loaded into memory.")

# Remove rows where 'price-usd' is zero

data = data[data['price-usd'] > 0]

# Normalize target (price)

data['price-usd'] = data['price-usd'] / 1e6 # Scale target to millions of dollars

# Prepare features and target

X = data[['bedrooms', 'bathrooms', 'sqft_living', 'sqft_lot', 'floors',

'waterfront', 'view', 'condition', 'sqft_above', 'sqft_basement', 'yr_built']].values

y = data['price-usd'].values

# Split the data

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Standardize the features

scaler = StandardScaler()

X_train = scaler.fit_transform(X_train)

X_test = scaler.transform(X_test)

# Define the ANN model

input_size = X_train.shape[1]

hidden_size = 32

output_size = 1

# Define the model with Input layer

model = tf.keras.Sequential([

tf.keras.layers.Input(shape=(input_size,)), # Explicit input shape

tf.keras.layers.Dense(hidden_size, activation='relu'),

tf.keras.layers.Dense(output_size)

])

# Compile the model

model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=0.0001),

loss='mse')

# Train the model

history = model.fit(X_train, y_train, epochs=1000, validation_data=(X_test, y_test), verbose=1)

# Plot training and validation loss

plt.figure(figsize=(10, 6))

plt.plot(history.history['loss'], label='Training Loss', marker='o')

plt.plot(history.history['val_loss'], label='Validation Loss', marker='x')

plt.title('Training and Validation Loss History')

plt.xlabel('Epoch')

plt.ylabel('Loss')

plt.legend()

plt.grid(True)

plt.show()

# Evaluate the model on the test dataset

y_pred_test = model.predict(X_test).flatten()

# Calculate accuracy for each prediction

accuracy = 100 * (1 - np.abs(y_pred_test - y_test) / y_test)

# Handle cases where actual price is zero (to avoid division by zero)

accuracy = np.where(y_test == 0, 0, accuracy)

# Create a DataFrame for predictions, actual prices, and accuracy

evaluation_df = pd.DataFrame({

'Predicted Price (Million $)': y_pred_test,

'Actual Price (Million $)': y_test,

'Accuracy (%)': accuracy

})

# Display first 10 rows of the table

print("\nEvaluation: Predicted vs Actual Prices with Accuracy")

print(evaluation_df.head(10))

# Save the evaluation table to a CSV file

evaluation_df.to_csv(r'd:\temp\evaluation_results_with_accuracy.csv', index=False)

# Plot Predicted vs Actual Prices

plt.figure(figsize=(10, 6))

plt.scatter(evaluation_df['Actual Price (Million $)'], evaluation_df['Predicted Price (Million $)'], alpha=0.7)

plt.xlabel('Actual Price (Million $)')

plt.ylabel('Predicted Price (Million $)')

plt.title('Predicted vs Actual Prices')

plt.plot([evaluation_df['Actual Price (Million $)'].min(), evaluation_df['Actual Price (Million $)'].max()],

[evaluation_df['Actual Price (Million $)'].min(), evaluation_df['Actual Price (Million $)'].max()],

color='red', linestyle='--', label='Perfect Prediction Line')

plt.legend()

plt.grid(True)

plt.show()

4.4 נתח את הקוד דרך CHATGPT איפה לדעתך יותר קל

טען את הקובץ ל CHATGPT בקש ממנו לנתח את הנתונים ולמצוא קשרים ולתת ניתוח מיללוי ברור על מגמה מובליה ואיפה הכי משתלם עבור השקעה

5. RUNWAY צפה בסרטון והעזר במדריך לפי הצורך

יסודות בינה מלאכותית : 10-RB27 – מבוא לרשת ניירונים

4.3 נוריד את הקובץ עבוד בינה מלאכותית מסוג ANN

USA Housing Dataset1

אולי תאהב/י גם

מכטרוניקה : רזוננס במחומרים ומערכות מכניות

ניטור מים – ובינה מלאכותית

פיתוח רחפן תכנון רחפן – חיבור בקר טיסה – החוטיידה

כתיבת תגובה לבטל