IIT M IMPROVEMENT AN EXAM QIP4 23 Feb - 23 Feb 2025

Concepts this paper tests

Auto-generated summary from the source site. Handy as a checklist; the lessons above explain each idea from scratch.

Summary of Core Topics, Concepts, Principles, and Equations

1. Machine Learning Workflow

• Train-Test Split: Essential to avoid overfitting and evaluate model performance on unseen data.

  • Risk of not performing train-test split: Model might overfit the training data and perform poorly on new data.

• Data Preprocessing: Necessary for cleaning and preparing data for model building.

  • Reasons for preprocessing:

    • Handling outliers or missing values.
    • Scaling features.
    • Converting non-numerical features to numerical representations.

2. Model Evaluation and Optimization

• Tolerance Parameter (tol): Used in optimization algorithms to specify the threshold for stopping criterion.

  • Equation: If the change in loss or parameters is less than tol, the optimization stops.
  • Example: SGDRegressor(tol=0.001)

• Cross-Validation: Techniques like Leave-One-Out Cross-Validation (LOOCV) for robust model evaluation.

  • LOOCV: Generates n models for n data points, leaving one out each time.

3. Data Transformation

• Polynomial Transformation: Used to create polynomial features.

  • Module: sklearn.preprocessing

• Label Encoding: Converting categorical data into numerical codes.

  • Example: LabelEncoder().fit_transform(data)

4. Evaluation Metrics

• Regression Metrics:

  • Mean Squared Error (MSE): $\text{MSE} = \frac{1}{n} \sum_{i=1}^{n} (y_i - \hat{y}_i)^2$
  • Mean Absolute Error (MAE): $\text{MAE} = \frac{1}{n} \sum_{i=1}^{n} |y_i - \hat{y}_i|$

5. Handling Missing Data

• Imputation: Techniques to handle missing values.

  • Example: SimpleImputer(strategy='median')

6. Statistical Measures

• Median: Used to replace missing values in categorical data.

  • Example: df["Contact"].median()

7. Data Manipulation and Analysis

• Pandas Operations: Techniques for data manipulation and analysis.

  • Example: df.groupby('Marital').agg({'Balance': 'median'})

8. Data Transformation Pipelines

• ColumnTransformer: Used to apply different transformations to different columns.

  • Example:

    transformers = [
        ('Ohe', OneHotEncoder(), [0, 2]),
        ('scaler', StandardScaler(), [1])
    ]
    ct = ColumnTransformer(transformers=transformers)