Hands-On Neural Networks​: Build Machine Learning Models​

Install Python 3.6.2 and the PyCharm Community Edition, verify the versions, configure the interpreter, and create a new project to run Python programs.

Learn to set up PyCharm, create and run Python files, customize the IDE, manage the project structure, and use the console, terminal, and run configurations.

Explore Python language basics by covering single value variables, including numbers and strings, and basic operations. Learn about commenting, printing, and dynamic typing in Python.

Explore multivalue variables in Python using tuples, arrays, and dictionaries, then practice indexing, slicing, and common operations like length, max, min, and insert, append, and remove.

Apply control flow concepts to code by using if, else if, and else statements, and while and for loops to test conditions, handle arrays, and implement nested logic.

Learn how to use functions in Python to organize code, pass parameters, return values, call functions, and reuse logic with practical examples like summing and multiplying arrays.

Explore how classes model objects with properties and behaviors in Python, create instances with initializers, and use inheritance, polymorphism, and method overriding to build complex game character hierarchies.

Explore the fundamentals of TensorFlow in Python, set up your first project, and learn core components and computational graphs for training a simple linear regression model.

Explore how TensorFlow provides high-level APIs and datasets to build, train, and evaluate machine learning models (e.g., linear regression) using tensors, graphs, and training–testing splits.

Learn to build a computational graph for a linear regression model with constant and operation nodes in TensorFlow, including importing the library and running sessions to output values.

Use placeholders as empty input nodes that gain a value when you run the session, supplying x values for models like linear regression to execute operations.

Understand how variable nodes differ from constant nodes, including initialization with a global variables initialiser and the ability to reassign values, plus the rich operations available on variable nodes.

Explore how to build a linear regression model from a computational graph by defining placeholders, variables, and constants, then train by minimizing loss to fit a line of best fit.

Build a linear regression model by constructing a computational graph with placeholders and variables, training with gradient descent to minimize loss and tune the slope and intercept.

Build a simple credit card fraud detection model with a neural network to classify transactions as fraudulent or legitimate, then train, test, and improve its accuracy using TensorFlow and Python.

Learn to build a basic credit card fraud detection model by preprocessing transactions, creating X/y train-test splits, and training a TensorFlow-based linear regression model to assess accuracy.

Explore a credit card transaction dataset with 28 features, a 0/1 fraud label, and train-test splits. Learn to import the data, build x_train and x_test, and apply pca.

Learn how to prepare a credit card fraud dataset by shuffling, one-hot encoding the class labels, normalizing features, and splitting into X_train, X_test, y_train, y_test (80/20) for ML modeling.

Balance imbalanced credit card data by applying logic racing to weight fraudulent transactions more heavily than legitimate ones. Learn ratio calculation, one-hot encoding, and bias reduction.

Build a multi-layer neural network by constructing a computational graph with inputs, outputs, and layers, then train by minimizing loss over epochs using placeholders, variables, and constants in TensorFlow.

Build a three-layer neural network to process training and testing data, using sigmoid and softmax activations, dropout, cross entropy loss, and Adam optimization, then measure accuracy.

Build and train a neural network by configuring layers, nodes, loss, and optimizer, then run a training loop with epochs, timing, and cross-entropy to monitor progress.

Test and evaluate your neural network by comparing final test accuracy to predictions, monitor the loss, and refine fraud-detection performance through iterative testing.

Build a simple stock market prediction model with TensorFlow to predict next-day price direction from volume data, using Python and investing.com resources for four stocks.

Build a simple neural networks model to predict whether a stock's next-day price rises or falls using end-of-day volume, with data exploration, csv datasets, and basic training.

Explore how to locate and download CSP stock datasets from investing.com, focusing on gold futures, and convert end prices, opening prices, and volumes to floats for training and testing.

Compute price differences by subtracting each day's final price from next day's opening price using final prices, opening prices, and volumes; train a model to predict differences from volume exchange.

Construct a simple regression model using a y = w x + b graph with placeholders, variables, a squared loss, and an optimizer, then train with a learning rate.

Create and train a neural network session by initializing global variables, setting epochs, and feeding training data to predict stock price differences, with later testing and accuracy evaluation.

Explore training a neural network, optimize loss by adjusting W and B, and test model accuracy by comparing predicted price differences against expected values using test volumes.

Import multiple stock data sets, create per-stock variables, and test a simple neural network predicting price differences from open prices and volume. Experiment with learning rate and epochs.