Objective:
Refresh knowledge of Python and mathematical principles.
Description:
Participants carry out analysis and data manipulation in Python
using the pandas and Matplotlib packages.
Chapter 1 – Data Analytics with Python:
Participants get to know our interactive programming
environment – the Data Lab – and brush up on key programming
and Python basics for data processing with Pandas, data
visualization with Matplotlib and Seaborn, and database querying
with SQL Alchemy.
Chapter 2 – Linear Algebra:
Participants become familiar with the mathematical background
of data science algorithms and learn the basic concepts of
linear algebra. Participants use the NumPy package to perform
calculations with vectors and matrices.
Chapter 3 – Probability Distributions:
Participants learn more about the statistical background of
data science algorithms. They explore important statistical
concepts and learn about discrete and continuous distributions.
Participants also gain an insight into code versioning with Git.

Objective:
Solving supervised and unsupervised machine learning problems
with sklearn.

Description:
Participants create data science workflows with sklearn, evaluate
their model performance using appropriate metrics, and learn about
the overfitting problem.

Chapter 1 – Supervised Learning: Regression:
Participants learn how to use the Python package sklearn with
linear regressions. They also learn about the regression model’s
assumptions and evaluating the predictions generated. Participants
learn about the bias-variance trade-off, regularization, and various
metrics of model quality.

Chapter 2 – Supervised Learning (Classification):
Participants are introduced to classification algorithms using the
k-Nearest-Neighbors algorithm and learn to evaluate the algorithm
and assess classification performance. They optimize the parameters
of their model and pay attention to dividing the data into training
and evaluation sets.

Chapter 3 – Unsupervised Learning (Clustering):
Participants learn about the k-Means algorithm as an example of
an unsupervised learning algorithm. They critically examine the
algorithm’s assumptions and performance metrics. Then they take a
brief look at an alternative to k-Means clustering.

Chapter 4 – Unsupervised Learning (Dimensionality Reduction):
Participants learn how to reduce the dimension of their data using
Principal Component Analysis (PCA) and use PCA to generate
uncorrelated features from the original data. In this context, they
explore the topic of feature engineering in more detail and new
features are generated from the old ones.

Chapter 5 – Outlier Detection:
Participants learn about different approaches to identifying outliers
and understand how to deal with these unusual data points. They use
robust measures and models to minimize the impact of outliers.

Objective:
Expanding the data science toolkit.
Description:
Participants deepen their knowledge of data classification models. In
doing so, they also expand their skills in collecting and preparing data.
Chapter 1 – Data Gathering:
Participants learn to gather data by mining web pages and PDF
documents. They structure collected text data using regular
expressions so that they can use it together with familiar algorithms.
Chapter 2 – Logistic Regression:
Participants learn a second classification algorithm: logistic regression.
They use new performance metrics to evaluate results and learn how
to prepare non-numeric data for their models.
Chapter 3 – Decision Trees and Random Forests:
Participants learn about the decision tree as an easy-to-interpret
model. They combine multiple models in an ensemble to improve
the predictions of their model. They also learn methods to deal with
unbalanced categories.
Chapter 4 – Support Vector Machines:
Participants learn about a final classification algorithm – Support
Vector Machines (SVMs) and examine the behavior of different kernels
for SVMs. They also learn the typical steps of Natural Language
Processing (NLP) and work through an NLP scenario using bag-ofwords models.
Chapter 5 – Neural Networks:
Participants are introduced to artificial neural networks and learn more
about deep learning, to create a multilayer artificial neural network
and apply it to a data set.

Objective:
Independent application of simple and complex modeling.
Description:
Participants gain confidence in solving data science problems and
learn to communicate results competently.
Chapter 1 – Visualization and Model Interpretation:
Participants learn important methods for interpreting and
visualizing machine learning models. By using model-agnostic
methods for interpretation, they learn to derive and communicate
insights into how their models work.
Chapter 2 – Spark:
Participants learn why working with distributed memory systems
is relevant. Using the Python package PySpark, they learn how
to read distributed databases, perform big data analysis, and use
well-known machine learning algorithms on distributed systems.
Chapter 3: Exercise Project:
Participants work on a prediction problem using a larger data set
and independently apply their data science skills from cleaning
the data set to interpreting the model. Participants receive
feedback on their approach to solving the problem in a project
consultation with StackFuel‘s mentoring team.
Chapter 4 – Final Project:
Participants are given another larger dataset to analyze
independently and solve with less assistance than they received
for the practice project. Participants receive feedback on their
solution approach in an individual project consultation with the
StackFuel mentoring team.