Machine Learning for Finance

Name: Machine Learning for Finance
Rating: 4.1 (65 reviews)

Machine Learning techniques for solving major financial issues

Created byPackt Publishing

Last updated 4/2020

English

What you'll learn

How to tackle problems in Fintech and financial investments
Learn feature engineering, EDA and understanding with regards to financial data
Build an ANN-based model for predicting the stock prices
Enhance your Machine Learning skills with ensemble models like random forest and XGBoost.
Enhance your understanding of Neural Networks to build regression-based models.
Learn how to identify fraudulent transactions by building a fraud detection model by using classification models.
Achieve efficient frontier by using features like Sharpe ratios and risk management.

Course content

7 sections • 49 lectures • 4h 30m total length

The Course Overview6:04
This video will give you an overview about the course.
Visualization, EDA, and Feature Engineering of Financial Data6:40
In this video, we will learn visualization, EDA, and feature engineering of financial data.
• Understand the concept of EDA
• Explore feature engineering of financial data
Features of the Stock Data4:32
In this video, we study about different features in the apple stock data and try to understand their meaning.
• Understand opening and closing price of stock
• Understand volume and other attributes
Univariate and Bivariate Analysis of Data14:28
In this video, we perform a univariate and bivariate analysis on the variables to understand their distribution, spread, and correlation.
• Look into the distribution and spread of variables
• Look into the evolution of price and volume
• Study the correlation between closing price and volume
Deriving Moving Average and RSI Based Features6:03
We derive the moving average, lag and RSI based features for the Apple stock data.
• Derive lag-based features
• Derive moving average and RSI based features
Data cleaning and Outlier Detection6:19
In this video, we clean the dataset by imputing null or empty values and detecting the outliers.
• Remove the null/empty values
• Detect and treat the outliers
• Study correlation to detect redundant variables
Creating the Features and Independent Variable6:37
In this video, we scrape the Wikipedia Apple page to get Apple events and add that as a predictor to enhance the feature space.
• Use beautiful soup to scrape the Apple data
• Add event as a Boolean variable to the stock data
• Study impact of apple events on its stock
Prepare Data for Modeling3:37
In this video, we normalize the independent variables and split them into training and test sets.
• Normalize the data
• Split the data into training and test sets and save it for further sections
Test your knowledge

Linear Regression Intuition5:41
In this video, we will understand how linear regression works and underlying assumptions of a linear model.
• What is linear regression
• How a linear model is formulated
• What are its assumptions
Understanding of FOREX Markets Data4:13
In this video, we will discover the FOREX markets in detail.
• Study about FOREX currencies
• Discuss about options and forward contracts
Pre-Process FOREX Currency Data for Model Input10:48
In this video, we’ll prepare the FOREX currency data (USD-JPY) for linear regression.
• Study the evolution of USD-JPY exchange rate over time
• Remove NA values and outliers
• Standardize the data for modelling
Building the Linear Regression Model6:54
In this video, we will build a linear regression model in order to forecast USD-JPY currency rates.
• Build a linear model using Statsmodel
• Measure model performance
R-Squared and Adjusted R-Squared as a Performance Metric4:18
In this video, we will learn about R-Squared and adjusted R-Squared, as well as, intuition behind them.
• Discover what R-Squared is
• Importance of having adjusted R-Squared
• Mathematical formulation of adjusted R-Squared
The Testing Significance of Features by Using p-value and VIF8:12
In this video, we will deep dive into p-values and VIF as feature selection mechanisms.
• Study p-value and intuition behind it
• Study VIF and write function to calculate VIF
Hyperparameter Tuning and Final Model Selection8:25
In this video, we use VIF and p-values to tune the model further and select best features.
• Drop features with high p-value and VIF
• Study best model performance
• Verify assumptions of linear regression
Test your knowledge

Decision Trees Intuition4:54
In this video, we will get an intuition about the working of decision trees.
• Study what a decision tree is
• How a decision tree works intuitively
Entropy and Information Gain Criterion for Tree Construction7:05
In this video, we’ll discuss about the two criterions for constructing decision trees which are information gain and entropy.
• See what entropy is and how it helps in tree construction
• See information gain as a tree construction criterion
Building a Decision Tree-Based Model for Predicting Stock Prices3:06
In this video, we build decision tree model for predicting the stock prices.
• Build a decision tree model with an initial set of parameters
• Visualize the tree
• Investigate the model performance
Train Using Different Max Depth4:13
In this video, we perform hyperparameter tuning of the decision tree model and train it with different max depths.
• Redefine the decision tree model with varying max depths
• Visualize and compare the model with previous ones
Random Forest Intuition3:22
In this video, we understand the idea behind ensemble models and study about random forest model in detail.
• Deep dive into ensemble models and why they are needed
• Study about types of ensemble models
• Deep dive into random forest model
Build a Random Forest Regressor for Predicting Stock Prices3:53
In this video, we build a random forest model to predict stock prices.
• Build a regression model using random forest
• Visualize importance of features
• Track the model performance
Boosting and XGBoost Based Regression Model for Stock Prediction4:33
We improve our stock prediction further by using XGBoost as the model.
• Build a XGBoost based regression model
• Compare model metrics with previous models
Test your knowledge

What a Neural Network Is8:39
In this video, we will introduce the neural networks, the building blocks, and its need.
• Define building blocks of a neural network
• See where it should be used
Feed Forward in Neural Networks4:31
In this video, we go over the process of feed forward in neural networks.
• Discuss about activation, neurons, and weights
• Study the flow of activations forward until the last layer
Gradient Descent in Neural Networks5:41
In this video, we study about gradient descent as an optimization technique and its applications in neural networks.
• Explain what gradient descent is
• Study its applications in neural networks
Back Propagation in Neural Networks7:52
In this video, we develop an in-depth understanding of back propagation and the mathematics behind it.
• See how neural networks are trained
• Intuition behind back propagation
• Math behind back propagation
Loss Function in Neural Networks4:08
In this video, we will look at the important loss functions in neural networks.
• Discuss about mean square error loss
• Discuss about sigmoid and log loss.
• Study about softmax and cross-entropy loss
Hyperparameters in Neural Networks7:59
In this video, we go to the Tensorflow playground to discuss different hyperparameters for neural networks.
• Explore different hyperparameters in neural network
• Tweak parameters on the playground
• Analyze how the result changes with change in hyperparameters
Test your knowledge

Prepare Data for Ingestion into the Neural Network3:26
In this video, we break the data into train and test pairs and calculate number of steps per epoch for the model.
• Divide data into train and test set
• Find number of steps per epoch
Define the Neural Network Layers and Model2:40
In this video, we define the architecture and the layers of the neural network
• Use Keras to define the neural network architecture
Visualize Keras Model by using Pydot2:52
In this video, we will visualize the regression model architecture by using Pydot library.
• Use Pydot to get intuitive understanding of architecture
Train the Model Using Basic Parameters2:12
In this video, we train out ANN model by using the basic hyperparameters.
• Define the basic parameters
• Train the model using basic parameters
Analyze the Model Performance Using Loss and Accuracy Curves1:33
In this video, we plot the training and validation loss to determine the optimum number of epochs.
• Plot the training and validation loss by using Matplotlib
Hyperparameter Tuning of Neural Network8:02
In this video, we use Talos library to do hyperparameter tuning of the neural network.
• Define hyperparameters to be scanned by using Talos
• Visualize Talos results by using line plot, scatter plot, and bar grid plot
• Determine the best combination of hyperparameters
Generating Predictions by Using the Trained Model2:47
In this video, we get the model with best hyperparameters and verify it on test data.
• Get best performing model from Talos
• Generate the predictions and calculate MAPE score
Test your knowledge

MPT and Stock Data Intuition8:12
In this video, we study about modern portfolio theory. Then, we retrieve stock data from Quandl and do a basic EDA on the stocks of Apple, Amazon, and Facebook.
• We start by studying what modern portfolio theory is
• Then we retrieve the stock data using Quandl
• Finally, we explore the evolution of Apple, Amazon, and Facebook stocks
Random Portfolio Generation and Portfolio Volatility7:19
In this video, we generate multiple random portfolio generation and study about their returns and volatility.
• Generate multiple random portfolio combinations
• Define functions to determine returns and volatility for portfolios
Sharpe Ratio for Optimum Portfolio3:44
In this video, we study about Sharpe ratio and how it helps to obtain an optimum portfolio.
• Study what Sharpe ratio is
• Define functions to determine Sharpe ratio for the portfolios
• Establish the mean returns and risk-free rate for the portfolios
Portfolio Allocation Using Sharpe Ratio and Efficient Frontier7:20
In this video, we simulate different portfolios and determine the optimum portfolio with maximum Sharpe ratio. We also study about efficient frontier.
• Define a method to simulate different portfolios and calculate Sharpe ratio and efficient frontier
• Determine best possible portfolio based on Sharpe ratio
Maximum Sharpe Ratio with SciPy Optimization4:56
In this video, we use the library SciPy to optimize the portfolio allocation.
• Define functions to maximize the Sharpe ratio using SciPy
• Define functions to minimize volatility using SciPy
Plotting and Visualizing Efficient Frontier8:33
In this video, we define functions to determine efficient returns for each portfolio and in turn visualize the efficient frontier.
• Define functions to calculate and plot the efficient frontier
• Visualize efficient frontier and portfolio allocation
Final Portfolio Allocation and Visualization6:25
In this video, we use the optimization functions defined in previous videos to devise the optimum portfolio allocation.
• Use optimization functions to determine optimum portfolio
• Visualize optimum portfolio and its allocation
Test your knowledge

Softmax and Sigmoid Activation in Neural Networks2:44
In this video, we study softmax and sigmoid activation for classification.
• Learn how softmax is used for multi-class classification
• Sigmoid activation for binary classification
Categorical Cross Entropy Loss for Classification4:00
In this video, we study categorical cross entropy and log-loss for multi-class and binary classification.
• Learn about categorical cross entropy for multiclass classification
• Understand log-loss for binary classification
Feature Engineering and Preprocess Data for Input into the Model4:56
In this video, we investigate the data and prepare it for modelling purpose.
• We start off with basic analysis of the credit card data
• We scale and split it into train and test sets
Creating the Model and the Optimizer2:24
We declare the artificial neural network model and compile it by adding the optimizer.
• Start off by declaring the neural network model
• Then define the metrics for model measurement
• Finally add the optimizer and compile the model
Training the Model5:06
We train the neural network model to measure loss, accuracy, and f1 score.
• Train the ANN model
• Plot training and test loss curve
• Look into confusion matrix and classification report
Handling Class Imbalance2:59
In this video, we fix the class imbalance by using weights for the classes.
• Define weights for the majority and minority class
• Define ANN model with new class weights
• Train the weighted ANN model
Evaluating the Final Model and Predict Fraud Using the Model5:27
In this video, we evaluate final weighted ANN model on the train and test data.
• Plot the training and test loss
• Find the optimal cutoff for probability threshold
• Look into confusion matrix and classification report
Test your knowledge

Requirements

Basic knowledge of Python, finance, and machine learning.

Description

Machine Learning for Finance is a perfect course for financial professionals entering the fintech domain. It shows how to solve some of the most common and pressing issues facing institutions in the financial industry, from retail banks to hedge funds.

This video course focuses on Machine Learning and covers a range of analysis tools, such as NumPy, Matplotlib, and Pandas. It is packed full of hands-on code simulating many of the problems and providing working solutions.

This course aims to build your confidence and the experience to go ahead and tackle real-life problems in financial analysis. The industry is adopting automatic, data-driven algorithms at a rapid pace, and Machine Learning for Finance gives you the skills you need to be at the forefront.

By the end of this course, you will be equipped with all the tools from the world of Finance, machine learning and deep learning essential for tackling all these pressing issues in the area of Fintech.

About the Author

Aryan Singh is a data scientist with a penchant for solving business problems across different domains by using machine learning and deep learning. He is an avid reader and has a keen interest in NLP research. He loves to participate and organize hackathons and has won a number of them. Currently, he works as a data scientist at Publicis Sapient.

Who this course is for:

This course is for financial professionals entering the field who already possess some Python skills and wish to become proficient in machine learning.

Machine Learning for Finance

What you'll learn

Explore related topics

Course content

Financial Data Understanding, EDA, and Feature Engineering8 lectures • 54min

Predicting the FOREX Currencies by Building a Linear Model7 lectures • 49min

Tree-Based Machine Learning Techniques for Stock Prediction7 lectures • 31min

Artificial Neural Networks Basics and Intuition6 lectures • 39min

Stock Price Prediction by Using Artificial Neural Networks7 lectures • 24min

Modern Portfolio Theory and Techniques for Portfolio Management7 lectures • 46min

Predicting Fraud in Financial Transactions by Using ANN classification7 lectures • 28min

Requirements

Description

Who this course is for: