RCS_ML_01_20

RCS Data Science and Machine Learning section January 2020 in conjuction with Accenture

Binder(cloud hosted Jupyter notebooks) Beta

Google Colab

- open all of current repo

Course Plan

Goal

Build a complete data analysis pipeline using Python ecosystem

• Define the problem
• Gather the raw data
• Process (clean) the data
• Explore
• Analysis (apply models, make predictions)
• Reports and Visual Results in a form understandable to stakeholders

Course Contents (50h)

Workplace Organization(~2h)

• Git version control / command line
• Jupyter / Anaconda environment for Data Science
• Text Editors

Python (~10h)

• Built in Data Types
• Control Structures
• Functions and Classes
• List/Dictionary Comprehensions
• File Manipulation
• Advanced Concepts (Generators/Decorators)
• useful Python standard libraries - Collections, functools, etc

external Numerical Libraries (~8h)

• NumPy/Pandas
• SciPy.Stats

noSQL databases noSQL: Hbase, MongoDB, Cassandra (~8h)

• principles, types, CAP
• Key-value DB, e.g., Redis
• Columnar db, e.g., HBase, Cassandra
• Document db, e.g., MongoDB
• Graph db, e.g., Neo4j [some practical tasks on each]

Projects

• get data, transform data

Machine Learning using Scikit-Learn, Keras(w/ Tensorflow), PyTorch ~12h

• Data Preperation - preprocessing, tidydata
• Training Data / Testing Data / splitting
• Supervised / Unsupervised learning
• Classification
• Clustering
• Regression
• Dimensionality reduction (curse of dimensionality)
• post-processing

Data Visualization Techniques - ~10h

• Visualization Libraries in Python, Plotly, matlplotlib
• Building your own dashboards with Flask web micro framework
• Dashboards with Tableau / PowerBi