Machine Learning using Python Programming

Learn the core differences between supervised and unsupervised learning, including labeled data for classification tasks like rain prediction and fruit type identification.

Learn the difference between supervised and unsupervised learning by visualizing labels, boundaries, and clustering based on data similarities; explore regression, classification, and association in practice.

This lecture explains the machine learning pipeline, with data collection as the first step. It shows that increasing data quantity and quality, along with labeling, improves model accuracy.

Apply data preprocessing to remove wrong values and outliers, and remove wrongly labeled examples to achieve uniform data points, helping the model learn patterns and improve accuracy.

Learn to select relevant inputs and engineer features for house price prediction, dropping irrelevant data like house names, and train models to predict prices on unseen data.

Understand machine learning terminology, including data sets, 70/30 data split for training and testing, model training and testing, prediction accuracy, and essential preprocessing for missing values.

Set up a Python machine learning environment with Anaconda and the Jupiter notebook, run notebooks locally via localhost, and organize code into runnable cells for modular experiments.

Create a two-dimensional matrix in Python by using nested lists and numpy, learn about a 3x3 matrix with three rows and columns, and apply zero-based indexing to access elements.

Learn to navigate two-dimensional numpy arrays by indexing and slicing rows and columns, including how to access individual elements, whole rows or columns, and understand array shapes.

Learn how to access a specific column in NumPy matrices by selecting a column index and using a colon to include all rows, yielding a column vector.

Explore NumPy statistical methods for machine learning, including sum, mean, min, max, and standard deviation, with examples on lists and two-dimensional arrays and converting lists to NumPy arrays.

Master NumPy matrix operations, including addition, multiplication and transpose, and create zero, one, and complex matrices, while exploring shapes, dtypes, and dot products.

Learn to iterate through a numpy array with nested for loops in Python, printing each element line by line, and upgrade from classical code to the standard library.

Understand Pandas, a core data science library, and its data structures—Series and DataFrame—where labeled one-dimensional series and multi-column dataframes enable data manipulation, analytics, and machine learning tasks.

Import pandas as pd and create a Series from a list, dictionary, or scalar value. Add optional labels and an index, using default zero-based labels that match the length.

Learn how to create a pandas series from a dictionary, with keys as the index and values as data, and how to update and handle missing indices.

Create a pandas series from a scalar by providing an index, so all positions share that value; the index argument is optional for some inputs like numbers or dictionaries.

Explore series processing in pandas by retrieving data from a dictionary, using labels and indices with square brackets to access series values for machine learning workflows.

create a data frame from a series by combining data from lists, dictionaries, or other frames; understand two-dimensional frames versus one-dimensional series, with optional index and columns.

Create a data frame from a list by building multiple named columns and setting an index that starts from one, illustrating how lists map to tabular data.

learn how to create a dataframe from a dictionary in Python, with keys as columns and lists as values; understand column naming and handling missing keys.

Learn how to directly create a pandas dataframe from numpy arrays, setting custom indices and columns. Construct dataframes from lists or dictionaries and understand the default indexing, starting from zero.

Learn how to load datasets with built-in functions, read Excel sheets and data frames, parse filenames, and access Kaggle and open-source datasets for analysis.

Create and inspect pandas data frames, use head, tail, and describe for statistics, and index with brackets to select rows and columns via the columns attribute.

Explore data visualization in Python using matplotlib and seaborn, with line plots, bar graphs, and scatterplots, including labeling and saving figures.

Learn to load the iris dataset from sklearn, explore four features and three classes, split data into train/test sets, and build a k-nearest neighbors classifier to measure accuracy.

Explore data preprocessing and binarization, converting values to binary with a threshold. See how class imbalance can bias a model and how balancing data, as in iris dataset example, helps.

Apply feature scaling by mapping data from its original range to a zero to one range using the dataset's minimum and maximum values, and transform data for consistent model performance.

Explore linear regression as a supervised learning algorithm, illustrating simple and multiple regression, fitting a line of best fit to predict unknown values by minimizing error.

Use gradient descent, the optimizer, to minimize the linear regression cost function, tuning weights and bias to fit data, whether in simple, multiple, or polynomial regression.

Understand gradient descent optimizer that minimizes the mean squared error cost function by updating weights beta with learning rate alpha and gradients to align predictions with actual values.

Apply simple linear regression to forecast demand from price data by computing the line of best fit from X (price) and Y (demand), including slope and intercept.

Build a linear regression model in Python using a salary versus experience dataset. Import, prepare X and y, plot the data, and split the data for training and evaluation.

Train a linear regression model on X and Y using 30 data points with an 80/20 train-test split, reshape data to two-dimensional form, and predict against observed values.

Visualize the line of best fit with matplotlib by training a linear regression model, plotting blue scatter points, and overlaying the red line to show predictions and extrapolation.

Explore why linear regression fails for binary classification and master logistic regression, which uses probabilities between zero and one to distinguish dog versus cat, with thresholding for final class.

Understand how the sigmoid function in logistic regression maps input to 0–1 and saturates at extreme values. See why its bounded output suits binary classification.

Explore how the confusion matrix assesses classification performance, detailing true positives, true negatives, false positives, and false negatives. Calculate accuracy for models like logistic regression.

Implement logistic regression in Python for binary classification, using standard scaling to normalize age and salary, split data into training and testing, train the model, and make predictions.

Learn to evaluate a classification model with a confusion matrix, identifying true positives, false positives, and false negatives, and visualize it as an annotated heatmap using Seaborn.

Learn how support vector machines extend logistic regression to improve classification and regression by maximizing the distance of the support vectors from the hyperplane.

Explore how support vector machines separate two classes by maximizing the margin with kernels, choosing linear boundaries or nonlinear options like polynomial based on the data.

Train a support vector classifier in Python, using SVC to classify data and compare with logistic regression, then evaluate predictions with a confusion matrix.