Predict fraud with data visualization & predictive modeling!

Set up your Python development environment by installing Python 3.6.2 and the PyCharm community edition, then create a new project and configure the interpreter.

Set up PyCharm, customize the environment, and create Python files. Navigate the project structure, use the console and terminal, and run configurations to execute Python programs.

Master Python basics by working with single-value variables, numbers and strings, and perform arithmetic and string operations: addition, modulus, concatenation, length, replace, split, and print through hands-on examples.

Master multivalue variables in Python, including tuples, arrays, and dictionaries. Learn indexing, slicing, and common operations like append, insert, delete on lists and multi-dimensional structures.

Master control flow by using if, else if, and else, along with while and for loops, to test conditions and drive code execution. Learn through traffic light examples and conditionals.

Learn how to define and call functions in Python, pass parameters, and return values to replace repetitive code and organize logic beyond control flow and loops.

Explore how classes represent objects with properties and behaviors, use initializers to create instances, and apply inheritance and method overriding to build capable subclasses in Python.

Explore the basics of TensorFlow with Python, and learn its core components and computational graphs. Build a simple linear regression model and set up a first project.

Learn how TensorFlow provides high-level APIs for building and training machine learning models using tensors and computation graphs, with datasets and cross-platform support from Python to web and mobile.

Explore constant and operation nodes in building a linear regression model with TensorFlow, learning how to create, run sessions, and combine values to form a computation graph.

Placeholder nodes hold no value until the session runs, then they receive input for the model, such as x values.

Explore variable nodes, their mutability, initialization, and their functions, contrasting them with constant nodes. Initialize variable nodes with a global initializer before use, reassign values with assign, and perform operations.

Build a regression model by constructing a computational graph with placeholders, variables, and constants, then train to minimize loss and fit a line of best fit.

Build and train a linear regression model with TensorFlow, using placeholders and variable nodes for x, y, W, and B, and minimize loss via gradient descent and an optimizer.

Explore the data set, build a simple credit card fraud detection model with Python and TensorFlow, and train and test it to predict fraudulent transactions.

Explore building a basic credit card fraud detection model by loading a dataset, preprocessing features, and training a TensorFlow linear regression model to evaluate accuracy on test data.

Explore a credit card transactions dataset to build fraud detection models using PCA-transformed features, balance skewed classes, and prepare training and testing data with time, amount, and 0/1 labels.

Learn to prepare and split a fraud detection dataset into four sets: x train, x test, y train, and y test, using shuffle, one-hot encoding, and normalization for predictive modeling.

Construct a multi-layer computational graph to separate legitimate from fraudulent data, specifying inputs, outputs, placeholders, variables, weights, and biases, and train by minimizing loss over multiple epochs.

Build a three-layer neural network to connect inputs through three layers, apply dropout and softmax, and optimize with cross-entropy and Adam to predict fraud from training and testing data.

Train a neural network with a three-layer graph, initializing variables, and run epoch-based optimization using cross-entropy and a feed dictionary fed by x_train and y_train for training.

Test the fraud detection model by running training and testing loops, evaluate accuracy and loss, and report final overall and fraud-specific accuracy on test data.

Learn an introduction to machine learning, four steps to build ML programs, and neural networks with a focus on convolutional neural networks via video lectures and text notes.

Explore what machine learning is, its core steps from data processing to training and testing a model, and how convolutional neural networks perform image classification under supervised learning.

Explore how neural networks use layers of interconnected nodes, weights, biases, and activation functions to map inputs to outputs and recognize patterns.

Explore convolutional neural networks by applying convolutions and max pooling to three-channel image data, turning pixel matrices into kernel-based features for dense layers.

Explore the Keras API to install and use Keras with TensorFlow libraries, build model components, train, evaluate, and predict, and learn activation functions and their advantages.

Introduce TensorFlow and Keras, explain their roles in machine learning, and guide installation and usage to build, train, and deploy models.

Explore Keras syntax with a sequential model, adding convolutional, max pooling, dense, and dropout layers, then compile, train, evaluate, and predict using gradient descent.

Explore activation functions such as identity, binary step, logistic sigmoid, tanh, ReLU, leaky ReLU, and softmax, and how they shape neuron outputs for convolutional neural network image classification.

Explore the CIFAR-10 dataset, format input and output data, and format input images and labels while coding along to load, display, and interpret model outputs.

Explore the Cifar-10 dataset by downloading and loading 32x32 color images across 10 classes, with 50k training and 10k testing samples, and learn to access images and labels in Python.

Understand how 32 by 32 color images are stored as RGB pixel matrices, with aligned train and test images and labels, then prepared via normalization and categorical encoding.

Format input images for a fraud prediction model by converting labels to one-hot encoding, applying softmax activation, normalizing pixel values, and flattening images for train/test data preparation.

Build our image classifier from scratch, train it, and study gradient descent and optimizers, with a hands-on coding day covering model construction, compiling, training, and evaluating results.

Build and train a model from scratch using convolutional layers, pooling, flattening, dense and dropout layers, with max-norm constraints and a softmax output for 10 categories.

Import and format inputs, set up train and test splits, and compile and train a model using a sagd optimizer, cross-entropy loss, and accuracy metrics.

Learn to save and load pre-trained models using the H5 format, evaluate and predict with them, and freeze and save graphs for mobile import in day seven.

Learn how to save and load a pre-trained model to an h5 file, including saving the model architecture and trained weights, and load it for evaluation and prediction without retraining.

Save a pre-trained model to a protobuf file by freezing the graph. Convert variables to constants for on-device deployment and enable Android and iOS imports.

Explore seven days of building an image classifier with Python and Karris, covering data formatting, convolutional neural networks, training, and saving trained models.