11/8/24, 3:48 PM concrete strength.

ipynb - Colab

# Import necessary libraries

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor
from xgboost import XGBRegressor
from sklearn.metrics import mean_absolute_error, r2_score, mean_squared_error
from sklearn.preprocessing import StandardScaler
from scipy.stats import zscore

# Load the dataset

data = pd.read_csv('/content/ConcreteStrengthData.csv')

# Display the first few rows of the data

print("Initial Data:\n", data.head())

Initial Data:
CementComponent BlastFurnaceSlag FlyAshComponent WaterComponent \
0 540.0 0.0 0.0 162.0
1 540.0 0.0 0.0 162.0
2 332.5 142.5 0.0 228.0
3 332.5 142.5 0.0 228.0
4 198.6 132.4 0.0 192.0

SuperplasticizerComponent CoarseAggregateComponent \
0 2.5 1040.0
1 2.5 1055.0
2 0.0 932.0
3 0.0 932.0
4 0.0 978.4

FineAggregateComponent AgeInDays Strength

0 676.0 28 79.99
1 676.0 28 61.89
2 594.0 270 40.27
3 594.0 365 41.05
4 825.5 360 44.30

# Step 1: Handle Missing Values

missing_value_strategy = input("Enter 'remove' to drop missing values or 'median' to fill
if missing_value_strategy == 'remove':
data = data.dropna()
print("Missing values removed.")
elif missing_value_strategy == 'median':
data = data.fillna(data.median())
print("Missing values filled with median.")

Enter 'remove' to drop missing values or 'median' to fill with median: median
Missing values filled with median.

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 1/7
11/8/24, 3:48 PM concrete strength.ipynb - Colab

# Display data after handling missing values

print("\nData after handling missing values:\n", data.head())

Data after handling missing values:

CementComponent BlastFurnaceSlag FlyAshComponent WaterComponent \
0 540.0 0.0 0.0 162.0
1 540.0 0.0 0.0 162.0
2 332.5 142.5 0.0 228.0
3 332.5 142.5 0.0 228.0
4 198.6 132.4 0.0 192.0

SuperplasticizerComponent CoarseAggregateComponent \
0 2.5 1040.0
1 2.5 1055.0
2 0.0 932.0
3 0.0 932.0
4 0.0 978.4

FineAggregateComponent AgeInDays Strength

0 676.0 28 79.99
1 676.0 28 61.89
2 594.0 270 40.27
3 594.0 365 41.05
4 825.5 360 44.30

# Step 2: Handle Outliers using IQR and Z-score

def remove_outliers_iqr(df):
Q1 = df.quantile(0.25)
Q3 = df.quantile(0.75)
IQR = Q3 - Q1
return df[~((df < (Q1 - 1.5 * IQR)) | (df > (Q3 + 1.5 * IQR))).any(axis=1)]

def remove_outliers_zscore(df, threshold=3):

z_scores = np.abs(zscore(df))
return df[(z_scores < threshold).all(axis=1)]

outlier_method = input("Enter 'iqr' for IQR method or 'zscore' for Z-score method to hand
if outlier_method == 'iqr':
data = remove_outliers_iqr(data)
print("Outliers handled using IQR.")
elif outlier_method == 'zscore':
data = remove_outliers_zscore(data)
print("Outliers handled using Z-score.")

Enter 'iqr' for IQR method or 'zscore' for Z-score method to handle outliers: zscore
Outliers handled using Z-score.

# Display data after handling outliers

print("\nData after handling outliers:\n", data.head())

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 2/7
11/8/24, 3:48 PM concrete strength.ipynb - Colab

Data after handling outliers:

CementComponent BlastFurnaceSlag FlyAshComponent WaterComponent \
0 540.0 0.0 0.0 162.0
1 540.0 0.0 0.0 162.0
5 266.0 114.0 0.0 228.0
7 380.0 95.0 0.0 228.0
8 266.0 114.0 0.0 228.0

SuperplasticizerComponent CoarseAggregateComponent \
0 2.5 1040.0
1 2.5 1055.0
5 0.0 932.0
7 0.0 932.0
8 0.0 932.0

FineAggregateComponent AgeInDays Strength

0 676.0 28 79.99
1 676.0 28 61.89
5 670.0 90 47.03
7 594.0 28 36.45
8 670.0 28 45.85

# Step 3: Normalize Data

scaler = StandardScaler()
scaled_data = scaler.fit_transform(data)
data = pd.DataFrame(scaled_data, columns=data.columns)
print("\nData after normalization:\n", data.head())

Data after normalization:

CementComponent BlastFurnaceSlag FlyAshComponent WaterComponent \
0 2.560014 -0.858514 -0.88112 -0.931973
1 2.560014 -0.858514 -0.88112 -0.931973
2 -0.112045 0.480231 -0.88112 2.346817
3 0.999687 0.257107 -0.88112 2.346817
4 -0.112045 0.480231 -0.88112 2.346817

SuperplasticizerComponent CoarseAggregateComponent \
0 -0.673726 0.839761
1 -0.673726 1.032749
2 -1.129625 -0.549747
3 -1.129625 -0.549747
4 -1.129625 -0.549747

FineAggregateComponent AgeInDays Strength

0 -1.288508 -0.229254 2.672454
1 -1.288508 -0.229254 1.590217
2 -1.365815 1.453139 0.701707
3 -2.345042 -0.229254 0.069106
4 -1.365815 -0.229254 0.631152

# Splitting the data into features and target

X = data.drop(columns=['Strength']) # 'Strength' is the target column
y = data['Strength']

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 3/7
11/8/24, 3:48 PM concrete strength.ipynb - Colab

# Train-test split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Verify the shapes of the split datasets

print("Training feature set shape:", X_train.shape)
print("Testing feature set shape:", X_test.shape)
print("Training target set shape:", y_train.shape)
print("Testing target set shape:", y_test.shape)

Training feature set shape: (784, 8)

Testing feature set shape: (197, 8)
Training target set shape: (784,)
Testing target set shape: (197,)

# Define Models
models = {
"Decision Tree": DecisionTreeRegressor(random_state=42),
"Random Forest": RandomForestRegressor(random_state=42),
"XGBoost": XGBRegressor(random_state=42, objective='reg:squarederror')
}

# Hyperparameter tuning for Random Forest and XGBoost using GridSearchCV

rf_param_grid = {
'n_estimators': [50, 100, 200],
'max_depth': [10, 20, None],
'min_samples_split': [2, 5, 10]
}

xgb_param_grid = {
'n_estimators': [50, 100, 200],
'max_depth': [3, 5, 10],
'learning_rate': [0.01, 0.1, 0.3]
}

# Tune and fit models

for model_name, model in models.items():
if model_name == "Random Forest":
grid_search = GridSearchCV(estimator=model, param_grid=rf_param_grid, cv=3, scori
grid_search.fit(X_train, y_train)
models[model_name] = grid_search.best_estimator_
print(f"\n{model_name} Best Parameters: {grid_search.best_params_}")
elif model_name == "XGBoost":
grid_search = GridSearchCV(estimator=model, param_grid=xgb_param_grid, cv=3, scor
grid_search.fit(X_train, y_train)
models[model_name] = grid_search.best_estimator_
print(f"\n{model_name} Best Parameters: {grid_search.best_params_}")
else:
model.fit(X_train, y_train)

Random Forest Best Parameters: {'max_depth': None, 'min_samples_split': 2, 'n_estimat

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 4/7
11/8/24, 3:48 PM concrete strength.ipynb - Colab
XGBoost Best Parameters: {'learning_rate': 0.3, 'max_depth': 5, 'n_estimators': 200}

# Model Evaluation Function

def evaluate_model(model, X_test, y_test):
predictions = model.predict(X_test)
mae = mean_absolute_error(y_test, predictions)
r2 = r2_score(y_test, predictions)
rmse = np.sqrt(mean_squared_error(y_test, predictions))
return mae, r2, rmse

# Evaluate each model

for model_name, model in models.items():
mae, r2, rmse = evaluate_model(model, X_test, y_test)
print(f"\n{model_name} Evaluation Metrics:")
print(f" MAE: {mae}\n R²: {r2}\n RMSE: {rmse}")

Decision Tree Evaluation Metrics:

MAE: 0.3109036376665524
R²: 0.7531568581871906
RMSE: 0.4824773835905145

Random Forest Evaluation Metrics:

MAE: 0.23158315377649288
R²: 0.8737109631559321
RMSE: 0.3451034120134297

XGBoost Evaluation Metrics:

MAE: 0.1958459295404801
R²: 0.8916473524633806
RMSE: 0.3196584515582839

# Delta Buckets Evaluation for Each Model

def delta_buckets(y_true, y_pred, delta=10):
differences = np.abs(y_true - y_pred)
within_delta = np.sum(differences <= delta) / len(differences)
return within_delta * 100 # Percentage of predictions within delta

# Display Delta Bucket Evaluation

for model_name, model in models.items():
y_pred = model.predict(X_test)
delta_10 = delta_buckets(y_test, y_pred, delta=10)
print(f"\n{model_name} - Percentage of predictions within ±10 units of true value: {d

Decision Tree - Percentage of predictions within ±10 units of true value: 100.0%

Random Forest - Percentage of predictions within ±10 units of true value: 100.0%

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 5/7
11/8/24, 3:48 PM concrete strength.ipynb - Colab
XGBoost - Percentage of predictions within ±10 units of true value: 100.0%

# Function to evaluate model performance

def evaluate_model(model, X_train, X_test, y_train, y_test):
model.fit(X_train, y_train)
predictions = model.predict(X_test)
mae = mean_absolute_error(y_test, predictions)
mse = mean_squared_error(y_test, predictions)
r2 = r2_score(y_test, predictions)
return mae, mse, r2

# Initialize models
models = {
"Decision Tree": DecisionTreeRegressor(random_state=42),
"Random Forest": RandomForestRegressor(n_estimators=100, random_state=42),
"XGBoost": XGBRegressor(n_estimators=100, random_state=42, objective='reg:squarederro
}

# Dictionary to store model performance

performance = {}

# Evaluate each model

for model_name, model in models.items():
mae, mse, r2 = evaluate_model(model, X_train, X_test, y_train, y_test)
performance[model_name] = {
"Mean Absolute Error": mae,
"Mean Squared Error": mse,
"R^2 Score": r2
}
print(f"{model_name} Performance:")
print(f"Mean Absolute Error: {mae:.2f}")
print(f"Mean Squared Error: {mse:.2f}")
print(f"R^2 Score: {r2:.2f}")
print("-" * 30)

# Display performance comparison

import pandas as pd
performance_df = pd.DataFrame(performance).T
print("\nPerformance Comparison:")
print(performance_df)

Decision Tree Performance:

Mean Absolute Error: 0.31
Mean Squared Error: 0.23
R^2 Score: 0.75
------------------------------
Random Forest Performance:
Mean Absolute Error: 0.23
Mean Squared Error: 0.12
R^2 Score: 0.87
------------------------------
XGBoost Performance:
Mean Absolute Error: 0.20
Mean Squared Error: 0.11
R^2 Score: 0.89

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 6/7
11/8/24, 3:48 PM concrete strength.ipynb - Colab
------------------------------

Performance Comparison:
Mean Absolute Error Mean Squared Error R^2 Score
Decision Tree 0.310904 0.232784 0.753157
Random Forest 0.230687 0.120433 0.872293
XGBoost 0.199964 0.107200 0.886325

https://colab.research.google.com/drive/1OaThaXg5S2tBw172V5NmebklkrofR9HK#scrollTo=HrAqhLmHxtqv&printMode=true 7/7