Master Automated Machine Learning :Build Real World Projects

Load the stroke prediction dataset, identify the 12th column as the target, and perform cleaning and visualization. Build and compare several classification models to maximize accuracy on train.csv and test.csv.

Prepare data, perform a train-test split, and compare gaussian naive bayes, decision tree, random forest, and neural networks on stroke prediction with test accuracies near 98%.

Apply PCA with three components to a stroke dataset, compare Gaussian NB, decision tree, random forest, and MLP classifiers, and assess performance with train-test splits and cross-validation.

import and explore a car price prediction dataset by loading pandas, numpy, and matplotlib; identify and encode categorical features: seller type, fuel type, transmission, and owner, and drop the car name.

Drop the car name, create a years-since-manufacture feature, apply one-hot encoding with drop-first to prevent the dummy variable trap, and build an extra trees model to predict selling price.

Compare random forest, XGBoost, CatBoost, and LightGBM regressors with tuning on a 70/30 split. Use mae, mse, rmse and feature engineering, highlighting XGBoost as best.

Analyze the Big Mart sales dataset with exploratory data analysis, apply log transformation to outlet sales, and preprocess categorical features using label encoding and one-hot encoding.

Explore building and evaluating multiple regression models to predict item outlet sales, dropping identifiers and using train-test split and cross-validation to compare linear, tree-based, and XGBoost approaches.

Import and explore a loan approval prediction dataset by loading it with pandas, numpy, seaborn, and matplotlib, describe attributes and data types, and identify the loan status as the target.

Explore data preprocessing and visualization for a loan dataset: fill nulls, apply log transforms, engineer total income, and label-encode categories for modeling.

Utilize randomized search CV to tune hyperparameters across random forest, decision tree, extra trees, XGBoost, and AdaBoost classifiers, improving accuracy by optimizing max depth, max features, and sample counts.

Import numpy, pandas, and matplotlib, and load a 25,491-row, 10-variable employee dataset to predict attrition, explore data, perform feature selection, fit models, and evaluate with ten-fold cross-validation for accuracy.

Apply recursive feature elimination to select the most predictive attributes for attrition, then compare logistic regression, random forest, SVM, and XGBoost using a 70/30 split and tenfold cross-validation.

Load numpy, pandas, matplotlib, and seaborn to build a hotel booking prediction pipeline with cross-validation, preprocessing, and model comparison using features like hotel type, lead time, arrival date, and cancelled.

Preprocess the hotel bookings data by cleaning missing values, replacing undefined entries, dropping zero-occupant rows, and saving a clean csv, then perform exploratory data analysis with plots and correlation matrix.

Perform data pre-processing and feature engineering, separating numerical and categorical features, handling missing values, and one-hot encoding; compare decision tree, random forest, logistic regression, and XGBoost with cross-validation.

Import libraries and load a dataset to build an apparent temperature prediction model. Evaluate performance with mean squared error and R2 score.

Compare linear regression, XGBoost, Lightgbm, CatBoost, and stacking regressors on weather features to predict apparent temperature; evaluate with MAE, MSE, RMSE, and stacking underperforms due to overfitting.

Assess how hyperparameter tuning improves regression performance across models—linear, CatBoost, XGBoost, and GBM—using mean absolute error, mean squared error, and RMSE on the wind dataset.

Import basic libraries and prepare a consumer complaint classification dataset by loading data, cleaning nulls, and encoding product categories with id mappings for model-ready training.

Examine category distribution and address imbalance with over sampling or down sampling; then convert text to fixed numerical vectors using tfidfvectorizer with sublinear tf, min_df=5, l2 norm, and 6081 features.

Convert text documents into numerical vectors with tf-idf and count vectorization, then compare svm, bernoulli naive bayes, random forest, xgboost, and gbm using pipelines and cross-validation.