Learn how to do forecasting using data science in 2 hours

Francisco J.

Play Speed
  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x
12 Videos (1h 50m)
    • Introduction

      4:09
    • General overview of time series

      6:24
    • Introduction to R

      12:35
    • Downloading data for our examples

      0:27
    • Very basic introduction to time series

      10:45
    • Arima manual

      15:23
    • First contact with auto.arima

      18:18
    • Introduction to auto.arima parameters

      7:20
    • auto.arima help and parameters

      8:59
    • Real example: predicting the US gdp

      16:03
    • Real example: predicting global temperatures

      7:21
    • Exercise: House prices in London

      2:36

About This Class

Imagine you want to predict sales, purchases, temperatures, or stock prices using a rigorous approach. Until recently, you had to be well versed in advanced mathematics and programming. Luckily, the tools have become cheaper and much easier to understand for the novice. You no longer need to be an seasoned data scientist or machine learning pro.

So, what can be forecasted? You can apply these techniques to any phenomenon you observe. Intuitively the predictions and models you generate will be "better" the more "structure" your phenomenon has. You cannot for example forecast the roulette results, because they are totally random! But you can probably forecast quite well your sales, salary, customers, expenses, website visitors, etc.

In this class, we will review how to use R (a leading free statistical software) for doing quite advanced forecasting, ignoring all the complicated statistical jargon. 

Statistical forecasting (aka time series analysis), is both a rigorous science and an art. Several experts, working with the same data, could potentially produce different forecasting models, each with its own pros/cons. Over the past years, there has been a plethora of automatic forecasting techniques which attempt to reduce the "human factor" of the approach (choosing one model among several ones). These techniques have gained a lot of popularity within companies doing forecasting on a large scale, such as banks, insurers, etc. But certainly, this has opened lots of possibilities for non-experts requiring to do forecasting.

After finishing this class, you should be able to: load data into R, produce a statistical forecasting model, validate the predictions and model accuracy, and work with those predictions.

This class will aim to teach the essential concepts behind these techniques, avoiding any mathematical and statistical concept, as much as possible. The list of topics will be roughly:

  1. Quick and direct introduction to R
  2. Reading data into R
  3. Essential introduction to time series and forecasting
  4. Forecasting in R. Leveraging the automatic methods
  5. Some forecasting
  6. Want to learn more? Some suggestions
  7. Exercise

8

Students

--

Projects

0

Reviews (0)