Python Machine Learning Step-by-Step: Modeling Financial Time Series Data

Transcription

1 Python Machine Learning Step-by-Step: Modeling Financial Time Series Data Reece Heineke Director of Big Data Credibly February 27, 2017

2 What is Machine Learning? Data Preparation Overview Python Toolbox Trade Ideas to Data Conclusion Exploratory Data Analysis Overview Scatter Plot Principal Component Analysis (PCA) Conclusion Fitting Models Overview Models and Pipelines Learning Curves Interpretability Conclusion A Fitted Model

3 What is Machine Learning?

4 What is Machine Learning? 1. Machine learning is a subfield of computer science that provides computers with the ability to learn without being explicitly programmed.

5 What is Machine Learning? 1. Machine learning is a subfield of computer science that provides computers with the ability to learn without being explicitly programmed. 2. There are two sides to every machine learning problem:

6 What is Machine Learning? 1. Machine learning is a subfield of computer science that provides computers with the ability to learn without being explicitly programmed. 2. There are two sides to every machine learning problem: 2.1 The learning

7 What is Machine Learning? 1. Machine learning is a subfield of computer science that provides computers with the ability to learn without being explicitly programmed. 2. There are two sides to every machine learning problem: 2.1 The learning 2.2 Model produced from the learning

8 Data Preparation: Overview Review the Python software stack

9 Data Preparation: Overview Review the Python software stack Motivate the problem

10 Data Preparation: Overview Review the Python software stack Motivate the problem Discuss some issues specific to time series modeling

11 Python Toolbox 1 1 Scientific Python by Eueung Mulyana

12 Trump2Cash 2 2 Trump2Cash GitHub Project

13 Input: Trump criticizes Toyota on Twitter

14 Output: Toyota stock opens lower 3 3 Toyota Stock on Yahoo Finance s Interactive Chart

15 WSJ Analysis of Trump Tweets 4 4 by Akane Otani and Shane Shifflett

16 IPython: A Data Scientist s Best Friend Jupyter Notebook

17 Data Preparation: Conclusion We now have a illustrative data set to work with Data set has 10 numeric dimensions: 9 inputs, 1 output

18 Data Preparation: Conclusion We now have a illustrative data set to work with Data set has 10 numeric dimensions: 9 inputs, 1 output Data set is large ( 400MB compressed)

19 Exploratory Data Analysis: Overview Covariance and Correlation Matrices

20 Exploratory Data Analysis: Overview Covariance and Correlation Matrices Scatter plots

21 Exploratory Data Analysis: Overview Covariance and Correlation Matrices Scatter plots Principal Component Analysis (PCA)

22 Exploratory Data Analysis: Overview Covariance and Correlation Matrices Scatter plots Principal Component Analysis (PCA) Kernel PCA

23 Using IPython Jupyter Notebook

24 Scatter Plot: What can we say about the data?

25 scikit-learn Algorithm Cheat-Sheet: Just looking 5 5 scikit-learn Cheat-Sheet

26 Principal Component Analysis (PCA)

27 Kernel PCA with Radial Basis Function (RBF)

28 Exploratory Data Analysis: Conclusion Nonlinear relationship with (0, 9), (2, 9), (6, 9)

29 Exploratory Data Analysis: Conclusion Nonlinear relationship with (0, 9), (2, 9), (6, 9) All other dimensions are quite random

30 Fitting Models: Overview Scikit learn s model and pipelines

31 Fitting Models: Overview Scikit learn s model and pipelines Illustrative learning curves

32 scikit-learn Revisited 6 6 scikit-learn Cheat-Sheet

33 scikit-learn Pipeline 7 7 Python Machine Learning by Sebastian Raschka

34 Holdout Method 8 8 Python Machine Learning by Sebastian Raschka

35 Cross-Validation 9 9 Python Machine Learning by Sebastian Raschka

36 Learning Curves: What does it tell us? Python Machine Learning by Sebastian Raschka

37 Poor fit: Linear Regression even with (K)PCA

38 Good fits: SVR (RBF) and Decision Tree Learning Curves

39 Classic Overfitting: Random Forest Regressor

40 Decision Trees: Easy to understand

41 Fitting Models: Conclusion Support Vector Machine (SVR) with Radial Basis Function (RBF) Kernel has a higher accuracy

42 Fitting Models: Conclusion Support Vector Machine (SVR) with Radial Basis Function (RBF) Kernel has a higher accuracy Decision Tree is easier to understand

43 Fitting Models: Conclusion Support Vector Machine (SVR) with Radial Basis Function (RBF) Kernel has a higher accuracy Decision Tree is easier to understand Choice involves our own priors on the underlying structure

44 Second Half of Machine Learning: A Persistent Model Jupyter Notebook

45 Thanks for listening: Q&A series modeling