​ ​ ​ AI Lab Manual

​ ​ ​ ​
​ ​ ​ ​ Session 2023-2027

​ ​ ​ Submitted By:
​ ​ ​ 2023-CS-156 Khola Raouf
​ ​ ​ Supervised By:
​ ​ ​ ​ Mr. Waseem

​ ​ ​ ​ ​ Course:
CSC-371 Artificial Intelligence

Department of Computer Science

University of Engineering and Technology
Lahore Pakistan
 2

Contents
Week # 9....................................................................................................................... 4
Question No. 1.............................................................................................................. 4
1.1.​ Importing Libraries:.......................................................................................... 4
1.2.​ Load the Dataset:.............................................................................................4
1.3.​ Data Preprocessing:.........................................................................................5
1.4.​ Model Initialization:...........................................................................................7
1.5.​ Model Prediction:............................................................................................. 7
1.6.​ Visualizing Results:.......................................................................................... 8
1.7.​ Evaluating the Model:.......................................................................................8
1.8.​ Final Output:.....................................................................................................9
1.9.​ Full Code:.......................................................................................................10
1.10.​ Model Scores:............................................................................................. 11
Question No. 2............................................................................................................ 13
2.1.​ Importing Libraries:............................................................................................13
2.2.​ Load Dataset:.................................................................................................... 13
2.3.​ Data Preprocessing:..........................................................................................14
2.4.​ Model Initialization:............................................................................................14
2.5.​ Model Training:...............................................................................................14
2.6.​ Model Prediction:...............................................................................................15
2.7.​ Visualizing Results:........................................................................................... 15
2.8.​ Model Evaluation:..............................................................................................16
2.9.​ Final Output:......................................................................................................17
2.10.​ Full code:........................................................................................................17
Question No. 03.......................................................................................................... 19
3.1.​ Evaluation Metrics for Classification Model:......................................................19
(a) Accuracy............................................................................................................ 19
(b) Precision............................................................................................................ 19
(c) Recall (Sensitivity or True Positive Rate)........................................................... 19
(d) F1-Score............................................................................................................ 19

2023-CS-156 ​
AI Tasks week#9
 3

(e) Confusion Matrix................................................................................................ 20

(f) ROC Curve & AUC (Area Under Curve)............................................................. 20
3.2.​ Evaluation Metrics for Regression Models........................................................ 20
(a) Mean Absolute Error (MAE)...............................................................................20
(b) Mean Squared Error (MSE)............................................................................... 20
(c) Root Mean Squared Error (RMSE).................................................................... 20
(d) R-squared (R2R^2 Score)..................................................................................20
3.3.​ Evaluation Metrics for Clustering Models.......................................................... 21
(a) Silhouette Score.................................................................................................21
(b) Davies-Bouldin Index......................................................................................... 21
(c) Dunn Index.........................................................................................................21
Conclusion...............................................................................................................21
Question No. 04..............................................................................................................22
4.​ Data Preprocessing Techniques........................................................................... 22
4.1.​ Handling Missing or Null Values.....................................................................22
4.2.​ Handling Duplicate Data................................................................................ 23
4.3.​ Handling Outliers............................................................................................23
4.4.​ Encoding Categorical Data............................................................................ 23
4.5.​ Feature Scaling (Normalization & Standardization)....................................... 24
4.6.​ Feature Selection & Engineering................................................................... 24
4.7.​ Splitting Data into Training and Testing Sets..................................................25
Final Summary Table.................................................................................................. 25

2023-CS-156 ​
AI Tasks week#9
 4

Week # 9
Question No. 1
Multiple Linear Regression
1.1.​ Importing Libraries:
First of all libraries are imported. Import
required Python libraries for handling data (pandas, numpy),
visualization (matplotlib, seaborn), and machine learning
(sklearn).
Code:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error,
r2_score
No output; libraries are just being imported.

1.2.​ Load the Dataset:

Load the dataset from a CSV file and display
the first few rows.
Code:
2023-CS-156 ​
AI Tasks week#9
 5

df = pd.read_csv("student_scores.csv") # Update with correct file

name
print(df.head()) # Display first few rows
A table will be showing the rows of the dataset, like:
Hours Previous Extracurricular Final Score
Studied Scores
5.1 78 2 82
6.3 85 3 88
4.8 75 1 80
7.2 90 4 92
5.5 79 2 83

1.3.​ Data Preprocessing:

1.3.1.​ Checking For Missing Values:
Ensure that there are no missing values in the
dataset.
CODE:
print(df.isnull().sum())
A summary of missing values per column, typically:

Hours_Studied ​ ​ ​ ​ 0
Previous_Scores ​ ​ ​ 0
Extracurricular ​ ​ ​ ​ 0
Final_Score ​ ​ ​ ​ 0

2023-CS-156 ​
AI Tasks week#9
 6

dtype: ​ ​ ​ ​ ​ ​ int64

If there were missing values, we would handle them using

methods like .fillna() or .dropna().

1.3.2.​ Checking For Data type:

Identify the data types of each
column to ensure compatibility with the machine learning model.
Code:
print(df.dtypes)

Hours_Studied ​ ​ ​ ​ float64
Previous_Scores ​ ​ ​ ​ int64
Extracurricular ​ ​ ​ ​ int64
Final_Score ​​ ​ ​ int64
dtype: ​ ​ ​ ​ ​ ​ object

If data types are incorrect, we may need to convert them using

.astype().

1.3.3.​ Select Features and Target Variable:

●​ X: Contains independent variables (features).
●​ y: Contains the dependent variable (Final_Score).
Code:
X = df[['Hours_Studied', 'Previous_Scores', 'Extracurricular']]
2023-CS-156 ​
AI Tasks week#9
 7

y = df['Final_Score']
No direct output, but X and y now store data for model training.

1.3.4.​ Split Data into Training and Testing Sets:

Splits data into 80% training data and 20% testing data.

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

No direct output, but now:

●​ X_train, y_train → Used for training

●​ X_test, y_test → Used for testing

1.4.​ Model Initialization:

Initialize a Linear Regression model and train it on
X_train and y_train.
​ ​ ​ Code:
model = LinearRegression()
model.fit(X_train, y_train)
No direct output, but the model has now learned the relationship
between features and Final_Score.

2023-CS-156 ​
AI Tasks week#9
 8

1.5.​ Model Prediction:

Use the trained model to make predictions on the
test
set.
​ ​ Code:
y_pred = model.predict(X_test)
No direct output, but y_pred contains predicted final scores for
X_test.

1.6.​ Visualizing Results:

●​ Compares actual vs. predicted values in a scatter plot.
●​ Ideally, the points should be close to a straight diagonal line.
Code:
plt.scatter(y_test, y_pred, color='blue')
plt.xlabel("Actual Scores")
plt.ylabel("Predicted Scores")
plt.title("Actual vs Predicted Scores")
plt.show()

A scatter plot showing how well the model's predictions match the
actual scores.

1.7.​ Evaluating the Model:

Evaluate model performance using:

2023-CS-156 ​
AI Tasks week#9
 9

●​ Mean Squared Error (MSE): Measures error magnitude (lower is

better).
●​ R² Score: Measures goodness of fit (closer to 1 is better)

Code:
mse = mean_squared_error(y_test, y_pred)
r2 = r2_score(y_test, y_pred)

print(f"Mean Squared Error: {mse}")

print(f"R-squared Score: {r2}")

1.8.​ Final Output:

2023-CS-156 ​
AI Tasks week#9
 10

1.9.​ Full Code:

import pandas as pd

import numpy as np

import matplotlib.pyplot as plt

import seaborn as sns

from sklearn.model_selection import train_test_split

from sklearn.linear_model import LinearRegression

from sklearn.metrics import mean_squared_error, r2_score

# Load dataset

df = pd.read_csv("Student_Performance.csv") # Ensure the correct file name

print("First few rows of the dataset:")

print(df.head()) # Display first few rows

# Check for missing values

print("\nMissing values in the dataset:")

print(df.isnull().sum())

# Check data types

print("\nData types of each column:")

print(df.dtypes)

# Convert categorical data to numerical (if needed)

df['Extracurricular Activities'] = df['Extracurricular Activities'].map({'Yes':

1, 'No': 0})

# Selecting features (independent variables) and target variable

X = df[['Hours Studied', 'Previous Scores', 'Extracurricular Activities', 'Sleep

Hours', 'Sample Question Papers Practiced']]

2023-CS-156 ​
AI Tasks week#9
 11

y = df['Performance Index']

# Split dataset into training (80%) and testing (20%) sets

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

random_state=42)

# Initialize and train the model

model = LinearRegression()

model.fit(X_train, y_train)

# Predict on test data

y_pred = model.predict(X_test)

# Visualizing results: Actual vs Predicted scores

plt.scatter(y_test, y_pred, color='blue')

plt.xlabel("Actual Performance Index")

plt.ylabel("Predicted Performance Index")

plt.title("Actual vs Predicted Performance Index")

plt.show()

# Model evaluation

mse = mean_squared_error(y_test, y_pred)

r2 = r2_score(y_test, y_pred)

print(f"\nMean Squared Error: {mse}")

print(f"R-squared Score: {r2}")

1.10.​ Model Scores:

PS E:\study\semester 4\AI\week 9> python task1.py
2023-CS-156 ​
AI Tasks week#9
 12

First few rows of the dataset:

Hours Studied Previous Scores Extracurricular Activities Sleep Hours Sample

Question Papers Practiced Performance Index

0 7 99 Yes 9 1 91.0

1 4 82 No 4 2 65.0

2 8 51 Yes 7 2 45.0

3 5 52 Yes 5 2 36.0

4 7 75 No 8 5 66.0

Missing values in the dataset:

Hours Studied 0

Previous Scores 0

Extracurricular Activities 0

Sleep Hours 0

Sample Question Papers Practiced 0

Performance Index 0

dtype: int64

Hours Studied int64

Previous Scores int64

Extracurricular Activities object

Sleep Hours int64

Sample Question Papers Practiced int64

Performance Index float64

dtype: object

2023-CS-156 ​
AI Tasks week#9
 13

Mean Squared Error: 4.082628398521853

R-squared Score: 0.9889832909573145

Question No. 2
2.​ Logistic Regression
2.1.​ Importing Libraries:
●​ Essential Python libraries for data handling, visualization, and machine learning
are imported.

Code:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import confusion_matrix, classification_report,
accuracy_score

Output:
No direct output, but the libraries are loaded successfully.

2.2.​ Load Dataset:

●​ Reads the dataset into a Pandas DataFrame and displays the first few rows.

Code:
df = pd.read_csv("student_logistic_regression.csv") # Update with actual file
name
print(df.head()) # Display first few rows

Output:
A table displaying the first five rows of the dataset, similar to:
User ID Gender Age EstimatedSalary Purchased
15624510 Male 19 19000 0

15810944 Male 35 20000 0

2023-CS-156 ​
AI Tasks week#9
 14

User ID Gender Age EstimatedSalary Purchased

15668575 Female 26 43000 0

15603246 Female 27 57000 0

15804002 Male 19 76000 0

2.3.​ Data Preprocessing:

●​ Drops unnecessary columns and converts categorical data into numerical values.
●​ Scales the numerical features for better model performance.

Code:
df = df.drop(columns=["User ID"]) # Remove irrelevant column
df = pd.get_dummies(df, columns=["Gender"], drop_first=True) # Convert Gender
to numeric
Output:
No direct output, but "User ID" is removed, and "Gender" is converted into a numeric column
(e.g., Gender_Male = 1 for Male, 0 for Female).

2.4.​ Model Initialization:

●​ Defines features (independent variables) and the target variable.
●​ Splits data into training and testing sets (80% training, 20% testing).

Code:
X = df.drop("Purchased", axis=1) # Features
y = df["Purchased"] # Target variable

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

random_state=42)
Output:
No direct output, but the dataset is split into training and testing sets.

2.5.​ Model Training:

●​ Standardizes feature values for improved performance.
●​ Initializes and trains the Logistic Regression model.

2023-CS-156 ​
AI Tasks week#9
 15

Code:
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)

model = LogisticRegression()
model.fit(X_train, y_train)
Output:
No direct output, but the Logistic Regression model is successfully trained on the dataset.

2.6.​ Model Prediction:

●​ Predicts outcomes using the trained model on the test dataset.

Code:
y_pred = model.predict(X_test)
Output:
No direct output, but y_pred contains the predicted values.

2.7.​ Visualizing Results:

●​ Compares actual vs. predicted values in a confusion matrix.
●​ The confusion matrix shows how well the model classified the test data.

Code:
cm = confusion_matrix(y_test, y_pred)
sns.heatmap(cm, annot=True, fmt='d', cmap="Blues")
plt.xlabel("Predicted")
plt.ylabel("Actual")
plt.title("Confusion Matrix")
plt.show()
Output:
A heatmap of the confusion matrix similar to:

Actual \ 0 1
Predicted
0 (No Purchase) 18 2

1 (Purchase) 3 7

2023-CS-156 ​
AI Tasks week#9
 16

2.8.​ Model Evaluation:

●​ Displays accuracy, precision, recall, and F1-score of the model.

Code:
print("Classification Report:\n", classification_report(y_test, y_pred))
print("Accuracy Score:", accuracy_score(y_test, y_pred))
Output:
A classification report and accuracy score, for example:
Classification Report:
precision recall f1-score support

0 0.86 0.90 0.88 20

1 0.78 0.70 0.74 10

accuracy 0.83 30
macro avg 0.82 0.80 0.81 30
weighted avg 0.83 0.83 0.83 30

Accuracy Score: 0.83

2023-CS-156 ​
AI Tasks week#9
 17

2.9.​ Final Output:

2.10.​ Full code:

3.​ # Importing Libraries
4.​ import pandas as pd
5.​ import numpy as np
6.​ import matplotlib.pyplot as plt
7.​ import seaborn as sns
8.​ from sklearn.model_selection import train_test_split
9.​ from sklearn.linear_model import LogisticRegression
10.​ from sklearn.preprocessing import StandardScaler
11.​ from sklearn.metrics import confusion_matrix, classification_report,
accuracy_score
12.​

2023-CS-156 ​
AI Tasks week#9
 18

13.​ # Load Dataset

14.​ df = pd.read_csv("Social_Network_Ads.csv")
15.​ print(df.head()) # Display first few rows
16.​
17.​ # Data Preprocessing
18.​ df = df.dropna() # Remove missing values
19.​ df = pd.get_dummies(df, drop_first=True) # Convert categorical variables to
numeric
20.​
21.​ # Define Features and Target
22.​ X = df.drop("Purchased", axis=1) # Replace with actual target column name
23.​ y = df["Purchased"]
24.​
25.​ # Split Data into Training and Testing Sets
26.​ X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,
random_state=42)
27.​
28.​ # Feature Scaling
29.​ scaler = StandardScaler()
30.​ X_train = scaler.fit_transform(X_train)
31.​ X_test = scaler.transform(X_test)
32.​
33.​ # Model Initialization and Training
34.​ model = LogisticRegression()
35.​ model.fit(X_train, y_train)
36.​
37.​ # Model Prediction
38.​ y_pred = model.predict(X_test)
39.​
40.​ # Evaluating the Model
41.​ print("Classification Report:\n", classification_report(y_test, y_pred))
42.​ print("Accuracy Score:", accuracy_score(y_test, y_pred))
43.​
44.​ # Confusion Matrix Visualization
45.​ cm = confusion_matrix(y_test, y_pred)
46.​ sns.heatmap(cm, annot=True, fmt='d', cmap="Blues")
47.​ plt.xlabel("Predicted")
48.​ plt.ylabel("Actual")
49.​ plt.title("Confusion Matrix")
50.​ plt.show()
51.​

2023-CS-156 ​
AI Tasks week#9
 19

Question No. 03
3.​ What are the evaluation matrices for a machine learning model?

Evaluation metrics are used to assess the performance of a machine learning model. The choice
of metric depends on the type of problem: classification, regression, or clustering.

3.1.​ Evaluation Metrics for Classification Model:

These metrics measure how well a model classifies data points into categories.

(a) Accuracy

●​ Measures the percentage of correctly predicted instances.

●​ Formula: Accuracy=Correct PredictionsTotal PredictionsAccuracy =
\frac{\text{Correct Predictions}}{\text{Total Predictions}}
●​ Best for: Balanced datasets (when classes are equally distributed).

(b) Precision

●​ Measures how many predicted positive instances are actually positive.

●​ Formula: Precision=True PositivesTrue Positives+False PositivesPrecision =
\frac{\text{True Positives}}{\text{True Positives} + \text{False Positives}}
●​ Best for: When false positives need to be minimized (e.g., spam detection).

(c) Recall (Sensitivity or True Positive Rate)

●​ Measures how many actual positive instances are correctly identified.

●​ Formula: Recall=True PositivesTrue Positives+False NegativesRecall =
\frac{\text{True Positives}}{\text{True Positives} + \text{False Negatives}}
●​ Best for: When false negatives need to be minimized (e.g., medical diagnoses).

(d) F1-Score

●​ Harmonic mean of Precision and Recall, providing a balance.

●​ Formula: F1=2×Precision×RecallPrecision+RecallF1 = 2 \times
\frac{\text{Precision} \times \text{Recall}}{\text{Precision} + \text{Recall}}
●​ Best for: Imbalanced datasets.

2023-CS-156 ​
AI Tasks week#9
 20

(e) Confusion Matrix

●​ A table that summarizes correct and incorrect predictions.

●​ Example:
Actual \ Positive (1) Negative (0)
Predicted
True Positive False Negative
Positive (1) (TP) (FN)

False Positive True Negative

Negative (0) (FP) (TN)

(f) ROC Curve & AUC (Area Under Curve)

●​ ROC Curve: Plots True Positive Rate (Recall) vs. False Positive Rate.
●​ AUC Score: Measures the area under the ROC curve (closer to 1 is better).
3.2.​ Evaluation Metrics for Regression Models
These metrics evaluate how well a model predicts continuous values.

(a) Mean Absolute Error (MAE)

●​ Measures the average absolute difference between actual and predicted values.
●​ Formula: MAE=1n∑i=1n∣yi−yi^∣MAE = \frac{1}{n} \sum_{i=1}^{n} | y_i - \hat{y_i} |
●​ Best for: When all errors are treated equally.
(b) Mean Squared Error (MSE)

●​ Measures the average squared difference between actual and predicted values.
●​ Formula: MSE=1n∑i=1n(yi−yi^)2MSE = \frac{1}{n} \sum_{i=1}^{n} (y_i -
\hat{y_i})^2
●​ Best for: Penalizing large errors more.

(c) Root Mean Squared Error (RMSE)

●​ Square root of MSE, providing an error in the same units as the target variable.
●​ Formula: RMSE=MSERMSE = \sqrt{MSE}
●​ Best for: Interpretable errors in real-world scenarios.

2023-CS-156 ​
AI Tasks week#9
 21

(d) R-squared (R2R^2 Score)

●​ Measures how well the model explains variance in the data (0 to 1).
●​ Formula: R2=1−∑(yi−yi^)2∑(yi−yˉ)2R^2 = 1 - \frac{\sum (y_i - \hat{y_i})^2}{\sum
(y_i - \bar{y})^2}
●​ Best for: Checking model goodness-of-fit (higher is better).

3.3.​ Evaluation Metrics for Clustering Models

These metrics evaluate unsupervised learning models like K-Means or DBSCAN.

(a) Silhouette Score

●​ Measures how well clusters are separated.

●​ Range: -1 to 1 (higher is better).
(b) Davies-Bouldin Index

●​ Measures the compactness and separation of clusters.

●​ Lower values indicate better clustering.
(c) Dunn Index

●​ Measures the ratio of minimum inter-cluster distance to maximum intra-cluster

distance.
●​ Higher values indicate better clustering.
Conclusion
Choosing the right evaluation metric depends on the problem type:

●​ Classification: Accuracy, Precision, Recall, F1-score, Confusion Matrix,

AUC-ROC
●​ Regression: MAE, MSE, RMSE, R-squared
●​ Clustering: Silhouette Score, Davies-Bouldin Index

2023-CS-156 ​
AI Tasks week#9
 22

2023-CS-156 ​
AI Tasks week#9
 23

Question No. 04
What are different methods or techniques used in Data Preprocessing?
(Like how we will handle the missing or null values in a data)

4.​ Data Preprocessing Techniques

Data preprocessing is a crucial step in machine learning that involves cleaning, transforming, and
preparing raw data for modeling. Below are the key methods used in data preprocessing:

4.1.​ Handling Missing or Null Values

Missing values in a dataset can cause issues in model training. There are several ways to handle
them:
(a) Removing Missing Values

●​ Method:
Drop rows or columns that contain missing values.
●​ Use When:The dataset is large, and missing values are minimal.
●​ Code Example (Pandas):
●​ df.dropna(inplace=True) # Removes rows with missing values
●​ df.drop(columns=['ColumnName'], inplace=True) # Removes a column with
missing values
(b) Imputing Missing Values

●​ Method: Fill missing values with statistical measures (mean, median, mode) or
interpolation.
●​ Use When:The dataset is small, and removing data is not ideal.
●​ Code Example:
●​ df['ColumnName'].fillna(df['ColumnName'].mean(), inplace=True) # Fill
with mean
●​ df['ColumnName'].fillna(df['ColumnName'].median(), inplace=True) # Fill
with median
●​ df['ColumnName'].fillna(df['ColumnName'].mode()[0], inplace=True) # Fill
with mode
(c) Using Machine Learning for Imputation

●​ Method: Predict missing values using K-Nearest Neighbors (KNN), Linear

Regression, etc.
●​ Code Example (KNN Imputer):
●​ from sklearn.impute import KNNImputer
●​ imputer = KNNImputer(n_neighbors=3)
●​ df[['Column1', 'Column2']] = imputer.fit_transform(df[['Column1',
'Column2']])

4.2.​ Handling Duplicate Data

Duplicate data can cause bias in the model.

2023-CS-156 ​
AI Tasks week#9
 24

●​ Remove Duplicates:
●​ df.drop_duplicates(inplace=True)

4.3.​ Handling Outliers

Outliers are extreme values that can distort model performance.
(a) Using the IQR (Interquartile Range) Method

●​ Removes values beyond 1.5 times the IQR range.

●​ Code Example:
●​ Q1 = df['ColumnName'].quantile(0.25)
●​ Q3 = df['ColumnName'].quantile(0.75)
●​ IQR = Q3 - Q1
●​ df = df[(df['ColumnName'] >= Q1 - 1.5 * IQR) & (df['ColumnName'] <= Q3 +
1.5 * IQR)]
(b) Using Z-Score Method

●​ Removes values that are more than a certain number of standard deviations
away.
●​ Code Example:
●​ from scipy import stats
●​ df = df[(np.abs(stats.zscore(df['ColumnName'])) < 3)]

4.4.​ Encoding Categorical Data

Categorical variables must be converted into numerical form for machine learning.
(a) Label Encoding (For Binary Categories)

●​ Converts categorical values into 0s and 1s.

●​ Example:
●​ from sklearn.preprocessing import LabelEncoder
●​ encoder = LabelEncoder()
●​ df['CategoryColumn'] = encoder.fit_transform(df['CategoryColumn'])
(b) One-Hot Encoding (For Multiple Categories)

●​ Creates separate binary columns for each category.

●​ Example:
●​ df = pd.get_dummies(df, columns=['CategoryColumn'], drop_first=True)
(c) Ordinal Encoding (For Ordered Categories)

●​ Assigns numerical values based on order (e.g., Low = 1, Medium = 2, High = 3).
●​ Example:
●​ from sklearn.preprocessing import OrdinalEncoder
●​ encoder = OrdinalEncoder(categories=[['Low', 'Medium', 'High']])
●​ df['CategoryColumn'] = encoder.fit_transform(df[['CategoryColumn']])

2023-CS-156 ​
AI Tasks week#9
 25

4.5.​ Feature Scaling (Normalization & Standardization)

Scaling is used to ensure that all features contribute equally to the model.
(a) Min-Max Scaling (Normalization)

●​ Scales values between 0 and 1.

●​ Use When:Data has no outliers.
●​ Example:
●​ from sklearn.preprocessing import MinMaxScaler
●​ scaler = MinMaxScaler()
●​ df[['Column1', 'Column2']] = scaler.fit_transform(df[['Column1',
'Column2']])
(b) Standardization (Z-Score Normalization)

●​ Scales values to have mean = 0 and standard deviation = 1.

●​ Use When:Data has outliers.
●​ Example:
●​ from sklearn.preprocessing import StandardScaler
●​ scaler = StandardScaler()
●​ df[['Column1', 'Column2']] = scaler.fit_transform(df[['Column1',
'Column2']])

4.6.​ Feature Selection & Engineering

Selecting the most important features improves model performance.
(a) Removing Irrelevant Features

●​ Drop unnecessary columns.

●​ df.drop(columns=['UnnecessaryColumn'], inplace=True)
(b) Using Correlation Matrix

●​ Removes features that are highly correlated (multicollinearity).

●​ Example:
●​ import seaborn as sns
●​ import matplotlib.pyplot as plt
●​ plt.figure(figsize=(10, 8))
●​ sns.heatmap(df.corr(), annot=True, cmap='coolwarm')
●​ plt.show()
(c) Using Feature Importance (Random Forest)

●​ Identifies the most important features.

●​ Example:
●​ from sklearn.ensemble import RandomForestClassifier
●​ model = RandomForestClassifier()
●​ model.fit(X, y)

2023-CS-156 ​
AI Tasks week#9
 26

●​ feature_importances = pd.Series(model.feature_importances_,
index=X.columns)
●​ feature_importances.nlargest(5).plot(kind='barh')

4.7.​ Splitting Data into Training and Testing Sets

To evaluate a model, data should be split into training and testing sets.
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,
random_state=42)

Final Summary Table

Preprocessing Step Methods
Handling Missing Values Drop missing values, Fill with mean/median/mode, KNN
Imputer

Handling Duplicates df.drop_duplicates()

Handling Outliers IQR method, Z-score method

Encoding Categorical Label Encoding, One-Hot Encoding, Ordinal Encoding

Data
Feature Scaling Min-Max Scaling, Standardization

Feature Selection Correlation matrix, Feature importance (Random Forest)

Data Splitting train_test_split() for training/testing

2023-CS-156 ​
AI Tasks week#9