Length 8 weeks, 48 hours
Weekly commitment Mon & Wed, 19:00 - 22:00
Period 9 July - 5 September, 2018
Location Sheung Wan, Hong Kong

Outcomes

1
Data mastery

Take data from any source, in a common format, and know how to transform it to become useful.

2
Machine learning

Intelligently pick up the best learning algorithm and back up that choice with evidence.

3
Business solutions

Take a business problem, and translate it into a multi-step data solution.

4
Communicate results

Communicate the most compelling insights a data model provides, quantify opportunities and potential risks.

Course Structure

Week 1
The Data Science Methodology

An insider's tour through to the world of data science and going hands on with its most popular tools.

  • Understand what data science is, how it relates to AI and how its employed to solve business problems.
  • Load in data from spreadsheets, text files, databases, and online sources.
  • Have mastery over data structures, transforming your datasets into the required formats and structures
Week 2
Exploratory Data Analysis

EDA is the practise of visually interrogating a dataset to understand what it'll be able to tell us.

  • Visually understand your data: What counts as an outlier? When do relationships look robust? What are data quality red flags?
  • Guide yourself to find out what's interesting about your dataset.
  • Recognise the cognitive biases people may have when looking at data through your graphs.
Project: TBC
Week 3
Statistical Data Analysis

A statistical tool-kit to ensure the validity of our data, and to evaluate the insights we may gain from it.

  • Be comfortable with hypotheses generation, testing, and evaluation. What are the questions that should always be asked?
  • Understand how to collect data using different sampling methods, and what the strengths and trade-offs of data sampling are.
  • Be aware of the limitations of your methods based on collected data.
Project: TBC
Week 4
Solving Estimation Problems

As an introduction to machine learning, we’ll solve a class of problems where you’d want to predict an amount, e.g. the popularity of content, the price at an auction, or the length of an incoming call.

  • Develop regression models (e.g. Linear, Ridge, LASSO, ElasticNet) and employ regularisation to avoid overfitting with high dimensional data.
  • Quantify the strength and confidence of your inferences.
  • Appreciate the trade-off between black-box and white-box models, and know when to use them.
Week 5
Solving Classification Problems

The second class of problems machine learning can solve is the separation of records and observations into the category they belongs to, e.g. whether a lead will convert, whether news is fake, or whether a market will go up or down.

  • Develop classification models (e.g. Random Forests, XGBoost, Neural Network) and tweak their performance through hyper parameter tuning.
  • Reason about business trade-offs by performing a cost/benefit analysis of your model’s various settings.
  • Evaluate your model’s performance with the confusion matrix.
Week 6
Time Series Analysis

Time is a special dimension. It’s inherently sequential and instead of just looking at a static snapshot, the momentum and curves along which change occurs can enter into our models.

  • Create a time series forecast based on trends and seasonality.
  • Perform pattern matching between real time data and historic records.
  • Cleanly aggregate (e.g. “on Friday’s we see X”) and resample (i.e. “the hourly trend shows Y”) time series data for descriptive insights.
Week 7
Feature Engineering

Data doesn’t just give up its secrets. That’s why data scientists engineer new features to separate the signal from the noise inherent in the data. Feature engineering is a proven strategy to lift your model’s performance.

  • Employ a mix of ingenuity and domain knowledge to construct novel and useful features.
  • Use unsupervised learning techniques like dimensionality reduction (e.g. PCA) and clustering (e.g. K-Means) to understand and describe the structure of your data
Week 8
Capstone Project

For the final project, you’ll bring everything you’ve learnt to bear on solving a problem of your choosing. Special emphasis is placed on the ‘last mile’ of Data Science - interpreting and presenting your results to a non-specialist audience.

  • Develop your own data science project from start to finish.
  • Successfully communicate important technical concepts to an audience with low data literacy using proven examples and illustrations.
  • Learn what happens after you’ve developed your model: making your model production ready, automatically refresh your models and online learning.
Project: Final Project

Instructor

Mart van de Ven Partner/Principal Data Scientist, Droste

Mart is the founding partner at a data science consultancy where he leads a team of data professionals on engagements with some of Hong Kong's most prestigious brands. He's a highly experienced and approachable instructor, having lead 1000+ hours of data science training and provided mentorship for 200+ capstone projects.

Is this course for me?

Data science foundations is aimed at professionals and graduates who want to:

  • Supercharge an existing skill-set (e.g. finance, marketing, software development) with a data science approach.
  • Carve out a niche for themselves in organisations which stand to benefit from adopting data-driven processes.
  • Gain an advantage over their peers through exposure to real-world challenges, data sets and best practices.
Prerequisites

In order to get the most out of the course, students should have a laptop (less than 4 years old) and be familiar with the following:

  • Python fundamentals - preparatory materials are available
  • Prior exposure to statistics - even if it was a lifetime ago
  • General computer literacy

Fees & Enrolment

Spectrum offers flexible payment options to ensure that nothing gets in the way of beginning your learning journey. From payment plans to earning your way through spectrum points - we've got your back. Please refer to our application page for full details on the enrolment process and payment options.


Data Science Foundations

9 July - 5 September
Mon & Wed, 19:00 - 22:00

HK$ 24,000 Total

Or give us a call to find out more:

+852 5808 2408