Linear Models and Time Series Forecasting

LEVEL: INTERMEDIATE

DURATION: 2-DAY COURSE

DELIVERED: AT YOUR OFFICE

What you will learn

You will learn the skills you need to develop and evaluate regression models that allow you to make quantitative predictions.

Multivariate linear regression

Autogression Models, ARMA, Prophet

Logistic Regression

Languages and libraries :

Python 3

Numpy and Pandas for data manipulation

Scikit-learn and statsmodel for linear and time series model

Matplotlib for visualisation

PREREQUISITES

Good knowledge of python, basic understanding of machine learning practice.

AUDIENCE

Those who wish to take their data science skills further and learn state-of-the-art techniques in this constantly evolving field.

Get in touch with us to learn more about the course!

Get in Touch

DAY ONE

Regression Analysis

Linear Regression
- Supervised learning
- Minimisation of residuals
- Mathematical formalism
- Implementing linear regression
Logistic Regression
- Classification problems
- The sigmoid function
- Loss functions
- Cross entropy
- Drawbacks of logistic regression
- Regularisation

DAY TWO

Time Series Analysis

Dealing with Time in Python
- Python datetime library
- Numpy date and time support
- Pandas date and time support
- Working with timezones
Introduction to Time Series
- Seasonality and trend
- Resampling
- Moving windows and moving average
- Differencing
- Autocorrelation
- Stationarity
Time Series: Autogression Models
Introduction to autoregression

ARMA
Time Series Testing
Cross validation with time-series
Prophet Library
Introduction to Facebook’s Prophet

Prophet in practice

Case Studies

The Role of Data Science in Retail – Argos

Learning Outcomes:

Get familiar with the theory of linear regression and how to implement the algorithm

Understand the theory behind logistic regression, how and when to use it in practice

Be able to work with time and date objects in Python, pandas and numpy

Learn how to process time-series data to prepare for analysis and forecasting

Learn the theory behind autoregression model and use ARMA in practice

Learn how to properly evaluate time-series models

Be able to use the Prophet library on time-series projects

KATE Projects:

Classification and Model Selection: Build a supervised learning model to predict the success of Kickstarter campaigns.
Time Series Analysis: Forecasting of daily energy consumption in London.

Get in Touch

Interested in learning more?

If you’re interested in what the ‘Linear Models and Time Series Forecasting’ module could do for your team or department, please complete the form to the right of this text and we’ll get back to you within two working days with more information.

Get in touch now

Please complete all of the required fields to get in touch with us. Alternatively, call +44(0)7816 419378 or email contact@cambridgespark.com now